SERVERS! BLADE CONFIGURATOR is a convenient tool that helps you quickly and easily choose the best option for server hardware for Blade systems.

BLADE servers are a unique technological solution. Data volumes are constantly growing, specialized Data Processing Centers or DPCs are intended for their processing and storage. But, if the volumes of Data are growing at a high speed, there is an urgent need for the density of the computers to be maximized, and such a density requires huge resources.

OBJECTIVE: To increase the number of computers located in a limited space, as well as reduce the cost of their maintenance.

Comparison of BLADE servers of different manufacturers

• BLADE server - in fact, it is an ordinary rack server only without a power supply and with passive radiators, when all "extra components" are removed;
• Actually - the case and passive Backplane board, providing switching of installed equipment;
• Power and cooling systems: fans and power supplies, taken out of the case;
• External switching devices: Ethernet, FC, Infiniband.

Instead of conventional PCI-E and PCI-X boards, mezzanine cards are inserted into the Blade Server, which allow the use of FC, Infiniband, SAS or additional Ethernet ports, provided that the corresponding switching module is available in the blade chassis.

MAIN ADVANTAGES OF BLADE SERVERS

The standard Blade server is an extremely compact server board that contains all the necessary components. Exactly the same ones exist on ordinary servers, with the only difference being that on “BLADES” both processors, memory, and network interfaces occupy a negligible amount of space. Thus, successfully coping with the same tasks, BLADE SERVERS, unlike their more bulky counterparts, really solve the problem of lack of free space for users. The small board is simply mounted in the chassis. This board commutes common power supplies, cooling system, common control system.

BENEFIT

• Reducing the total cost of IT infrastructure costs;
• Ensuring the reliability of power and cooling systems;
• Reducing the number of external ports, patch wires, as well as power cables;
• Improving the convenience of managing the system as a whole;
• Decrease in the occupied area;
• reduction in energy consumption and heat dissipation;
• High scalability
• Flexibility!

Another rather significant advantage of BLADS is this: after the introduction of BLADE technology, the standard DATACENTER will not only decrease in size, thanks to the use of BLADE servers, modern IT departments can get rid of the use of expensive patented equipment and software, that is, server management will not only more open, but also simpler. And this is an excellent reason to pay less for renting the data center premises. An important advantage is the fact that the task of managing a large number of server “blades” is quite capable of being handled by only one technical specialist!

In what cases should I radically change IT infrastructure and switch to the implementation of BLADE SERVERS?

In its most general form, the main motive for the transition to the procurement of Blade systems is the general MODERNIZATION of SERVER INFRASTRUCTURE and the desire of customers to move from an existing server zoo to a more modern and streamlined server system in the organization.

An important additional argument in favor of choosing Blade systems is the availability of a real or planned SOLUTION FOR CONSOLIDATING DATA STORAGE SYSTEMS. In favor of this consideration, one can also cite the fact that requests for Blade server specifications very often involve ordering Disk arrays of data, for example, HP EVA or HP MSA. Indeed, the simultaneous transition to BLADE and consolidated data storage has an undeniable advantage, because you no longer need to store large amounts of information in the external memory of the servers themselves! In all other respects, modern Blade Servers are not inferior to their rack-mount counterparts.

Using Blade servers to organize a terminal system: the choice of Blade for implementing a Server-Terminal system is very common. The advantages and conveniences of the server construct include the absence of a large number of connecting cables for the shelves, convenient visualization of the shelves, and quick identification of individual modules in it.

Properly selected BLADE equipment will not only ensure stable operation of the system, but will also help save money on the acquisition and maintenance of server infrastructure in the future!

The turn to professionals in the field of design and implementation of BLADE technologies - ADVANSERV.

Completed projects

Quick server selection

Questions and answers

Key benefits of blade servers: less rack space, higher density, more efficient cooling and more energy efficient, simplified administration

The main advantages of blade servers: less rack space, higher density, more efficient cooling and more energy efficient, simplified administration of a large number of servers. Additionally, blade servers can be more cost-effective if you need to connect to LAN and SAN at high speeds. Accordingly, if these problems are relevant to you, it is worth considering the possibility of acquiring a blade server.

Almost all DBMSs have built-in backup devices. If you need advanced functionality, you must use a third-party backup tool

Almost all DBMSs have built-in backup devices. If you need advanced functionality, you must use a third-party backup tool. It can be both free software and paid. Depending on the software licensing model, you need to purchase a specific set of licenses.

Fault tolerance refers to the ability of a system to work when one or more of its components fails. Accordingly, to ensure fault tolerance

Fault tolerance refers to the ability of a system to work when one or more of its components fails. Accordingly, to ensure the fault tolerance of the system, all its components must be duplicated. In each case, this list will vary depending on the deployed solution. You can also make a list of failures (threats) from which you need to protect the system, for example: a power supply failure, failure of RAM, the entire server, one of the services on which the application depends, the UPS of the rack, switch, server room as a whole.

A cluster is a group of computers connected by high-speed communication channels and representing a single hardware resource from the point of view of the user. The cluster serves

Cluster   - a group of computers connected by high-speed communication channels and representing a single hardware resource from the point of view of the user. The cluster serves to distribute the load, to increase fault tolerance. To create a cluster, you need software that supports this function, two or more servers, and, in most cases, a data storage system (SHD). Data Storage System (SHD) is a comprehensive software and hardware solution for organizing reliable storage of information resources and providing guaranteed access to them. Examples of cluster solutions:

• Stratus
• VMware vSphere
• Oracle RAC
• MS NLB, Failover cluster
• Symantec Veritas Cluster Server

3C is involved in the work of 1C - Client / Server 1C / DBMS Server. You must monitor all nodes using the standard Windows Server monitoring tool, perfmon. Necessary

3C is involved in the work of 1C - Client / Server 1C / DBMS Server. You must monitor all nodes using the standard Windows Server monitoring tool, perfmon. It is necessary to track indicators: processor load, free RAM, number of page read errors, disk queue for each disk, read / write speed to disk, network card queue.

A variety of devices can be used as a means for storing backups, from USB Flash to tape libraries and disk arrays. Range of solutions is enough

A variety of devices can be used as a means for storing backups, from USB Flash to tape libraries and disk arrays. The range of solutions is wide enough, as is the range of requirements, so you cannot give a definite answer. In any case, it is necessary to determine the key indicators: The volume of backups, the number of backups stored, the frequency of their execution, the backup window, the necessary recovery time, the necessary speed of backups and recovery, the need for replication or movement of backup media. In general, for medium-sized companies (SMBs) which we recommend using backup to the HDD, or to the NAS, or to a specialized device with deduplication, because of simpler administration and shorter recovery time. If there are specific requirements, then we are ready to individually select a solution.

A typical server uses only 5-10% of its capabilities. Virtualization allows you to combine the resources of the processor, memory and drives of hardware servers and provide

A typical server uses only 5-10% of its capabilities. Virtualization allows you to combine the resources of the processor, memory and drives of the hardware servers and provide shared resources for many virtual servers (virtual machines), efficiently and simply manage and monitor a significantly smaller number of hardware servers, and deploy many virtual ones ten times faster. This reduces the cost of electricity and cooling, reduces the occupied space, cable infrastructure.

Creating a RAID RAID array on a server To create a RAID array on a server, you must first have HDDs connected to the server itself. Motherboard,

Creating a RAID RAID array on the server

To create a RAID array on a server, it is necessary, first of all, to have HDDs connected to the server itself. The motherboard installed in the server must be either with an integrated RAID controller (built into the motherboard), or you will need to install a separate discrete RAID controller, which, as a rule, is installed in a special PCI-Express slot.

RAID Level Comparison

* N - the number of disks in the array, S - the size of the smallest disk.
  ** Information will not be lost if the drives fail within different mirrors.
  *** Information will not be lost if the same number of disks in different stripe "ah fails.
  **** Information will not be lost if the drives fail within the same mirror.

With components taken out and generalized in a basket to reduce the occupied space. Basket   - A chassis for blade servers that gives them access to common components, such as power supplies and network controllers. Blade servers are also called ultra-compact servers.

Internal structure

In the blade server, some typical components traditionally present on the computer are missing or out. The functions of power supply, cooling, network connection, connection of hard drives, server-to-server connections and control can be assigned to external units. Together with them, a set of servers forms the so-called blade system.

The processor and RAM must be placed in the server blades, the remaining components can be put into the basket in principle; the concept of a blade server provides for the replacement of part of the remaining components by external units (power supplies) or their virtualization (input / output ports, control consoles), thereby greatly simplifying and simplifying the server itself.

External Pluggable Units

Blade systems consist of a set of blade servers and external components that provide non-computing functions. As a rule, components that create a lot of heat, take up a lot of space, and are also repeated in functions between servers, are taken out of the server motherboard. Their resources can be distributed between the entire set of servers. The division into internal and external functions varies from one manufacturer to another.

Power supplies

A power supply converter is typically created common to the entire blade system. It can be either mounted inside it, or taken out in a separate unit. Compared to the sum of the individual power supplies required by the format servers, a single power supply for blade systems is one of the most significant sources of space saving, power consumption and the number of electronic components.

Cooling

The traditional server design attempts to balance the density of electronic components and the ability to circulate cooling air between them. In blade designs, the number of protruding and large parts is minimized, which improves cooling of the modules.

Network connections

Modern network interfaces are designed for extremely high data rates through conductive and optical cables. Such equipment is expensive and takes place in the design of the server. A frequent case is the excessive bandwidth of network interfaces, whose capabilities are not in demand in practical tasks. Combining network interfaces into a single device or using special blade slots that are exclusively dedicated to working with the network can reduce the number of connectors and reduce the cost of each connection.

Using disk drives

Although significant capacities are required to store volumes of data and programs, they do not need to be located locally. Interfaces such as FireWire, SATA, SCSI, DAS, Fiber Channel and iSCSI allow you to connect drives at a considerable distance from the processors. By analogy with network connections (and the iSCSI interface relies only on them), the corresponding devices can be placed in the enclosure of the blade system or mounted on dedicated blade slots.

A special solution in the form of a blade system loaded through a storage area network (SAN) allows you to create an extremely reliable and compact server system.

Specialized Blade Slots

Standardization of the blade slot interface allows you to create devices that can not only perform calculations, but also provide other services, for example, the functions of a network switch, router, quick connection to a local network or fiber. These resources can be used by other blade slots.

Areas of use

In standard server racks, the minimum server size is 1 unit, as a rule, such racks can accommodate 42 units of equipment, that is, a maximum of 42 servers when placed without baskets. Using blade servers allows you to circumvent this limitation without going beyond the size of a standard rack and place up to 100 servers in each.

Blade servers are especially effective for solving specific tasks: web hosting, cluster organization. Rack servers typically support hot swapping.

Although the technology for building blade systems is not proprietary (owned by one company), using components from the same manufacturer has fewer problems with installation and configuration. Standardization of interfaces could make the technology more accessible to the user and expand the choice of suppliers.

For all its merits, this technology cannot be considered the solution to all server problems. Large tasks still require the use of larger systems to solve them, such as mainframes and clusters. Clusters consisting of blade servers can also be used. Such a structure is particularly susceptible to the problem of overheating due to the dense layout of the electronics in each of them.

History of creation

Microcomputers that fit in 1 slot of a standard 19-inch rack began to be used as early as the 1970s, shortly after the development of 8-bit microprocessors. Sets of such slots were used in the management of industrial facilities, replacing mini-computers. Programs were recorded in an electrically programmable memory (EPROM) on a slot; their functionality was limited to one simple operation performed in real time.

The name "blade server" has been applied to systems with installed hard drives or flash memory. Thanks to this, it became possible to execute a full-fledged operating system on the slot.

Technology borrows some of the mainframe features. However, it is more correct to consider a group of blade slots as a cluster of independent servers, and the active use of resource virtualization and tight integration with the operating system are possible, which further increases productivity and stability.

The first manufacturer of blade systems is considered RLX Technologies (Houston, USA), founded by immigrants from Compaq. The company was bought by Hewlett-Packard in 2005.

As of the second half of the 2000s, the leader in the production of blade systems was Hewlett-Packard, whose share in the international market for 2008 in terms of revenue from sales and the number of pieces of equipment manufactured and delivered was 54.0% and 50.6, respectively %, which is double the share of company No. 2, IBM (23.9% and 25.9%). Other major market participants: Dell (global market shares of 8.1% and 9.5%, respectively), Fujitsu Siemens Computers (3.5% and 2.1%), Sun (4.1% and 4.0%) , Hitachi (1.0% and 0.6%) and NEC (1.6% / 1.2%).

Other companies, such as Appro, Intel, Rackable (SGI), Verari, Supermicro, also produce blade servers with varying degrees of success, but they occupy a few tenths of a percent or even less in the blade server market.

Write a review on the article "Blade Server"

References

see also

Excerpt about Blade Server

The living room of Anna Pavlovna began to fill up a little. The highest nobility of St. Petersburg arrived, people are the most heterogeneous in age and character, but the same in society, in which everyone lived; the daughter of Prince Vasily, the beautiful Helene, came to call on her father to go with him on the feast of the messenger. She was in a cipher and a ball gown. Arrived also known as la femme la plus seduisante de Petersbourg [the most charming woman in St. Petersburg,], the young, little princess Bolkonskaya, who married last winter and now did not go out into the light because of her pregnancy, but went on small evenings. Prince Hippolytus, son of Prince Basil, arrived with Mortemar, whom he introduced; Abbot Morio and many others arrived.
  - You have not seen yet? or: - you are not familiar with ma tante [with my aunt]? Said Anna Pavlovna to the guests who arrived and took them very seriously to a little old woman in tall bows, who had come up from another room, as soon as the guests began to arrive, called them by name, slowly shifting her eyes from the guest to ma tante [aunt], and then left.
  All guests performed a ceremony of greeting to an unknown, uninteresting and unnecessary aunt. Anna Pavlovna with sad, solemn participation watched their greetings, tacitly endorsing them. Ma tante spoke to everyone in the same terms about his health, about his health and about the health of her Majesty, which today, thank God, is better. All who approached, without decency showing no haste, with a sense of relief of a heavy duty, departed from the old woman so that she would never come to her all evening.
  The young princess Bolkonskaya arrived with work in an embroidered golden velvet bag. Her pretty, with a slightly blackened tendril upper lip was short in the teeth, but the milder it opened and the milder it stretched sometimes and fell to the lower one. As is always the case with quite attractive women, her flaw - the shortness of her lips and half-open mouth - seemed to her especially, in fact, her beauty. It was fun for everyone to look at this, full of health and liveliness, a pretty expectant mother, so easily enduring her position. It seemed to the old people and the bored, gloomy young people who looked at her that they themselves were becoming like her, having been and talked for a while with her. Anyone who spoke to her and saw at every word her bright smile and brilliant white teeth that could be seen without ceasing, he thought that he was especially kind today. And everyone thought that.
The little princess, waddling around, walked around the table with her quick handbag in her small quick steps and sat cheerfully straightening her dress, sat on the sofa, near the silver samovar, as if all she did was part de plaisir [entertainment] for her and for everyone her surrounding.
  “J” ai apporte mon ouvrage [I took over the work], ”she said, unfolding her reticule and addressing everyone together.
  “Look, Annette, ne me jouez pas un mauvais tour,” she said to the hostess. - Vous m "avez ecrit, que c" etait une toute petite soiree; voyez, comme je suis attifee. [Don't play a bad joke with me; you wrote to me that you have a very small evening. See how I am dressed badly.]
  And she spread her arms to show her, in lace, a gray graceful dress, belted a wide ribbon a little below her breasts.
  “Soyez tranquille, Lise, vous serez toujours la plus jolie [Be calm, you will all be the best],” answered Anna Pavlovna.
  “Vous savez, mon mari m" abandonne, ”she continued in the same tone, addressing the general,“ il va se faire tuer. Dites moi, pourquoi cette vilaine guerre, [You know, my husband leaves me. He is going to die. Tell me , why this nasty war,] she said to Prince Vasily and, without waiting for an answer, turned to the daughter of Prince Vasily, to beautiful Helen.
  - Quelle delicieuse personne, que cette petite princesse! [What a lovely princess this little princess is!] Said Prince Vasily quietly to Anna Pavlovna.
  Soon after the little princess came a massive, fat young man with a cropped head, glasses, light trousers in the fashion of that time, with a tall frill and a brown tailcoat. This fat young man was the illegitimate son of the famous Catherine nobleman, Count Bezukhov, who was now dying in Moscow. He had not served anywhere else, had just come from abroad, where he was brought up, and was for the first time in society. Anna Pavlovna greeted him with a bow relating to the people of the lowest hierarchy in her salon. But, despite this inferior greeting, the sight of Pierre in the person of Anna Pavlovna portrayed anxiety and fear, similar to that expressed by the sight of something too vast and unusual for the place. Although, indeed, Pierre was somewhat larger than the other men in the room, this fear could only apply to that clever and at the same time timid, observant and natural look that distinguished him from everyone in this living room.
- C "est bien aimable a vous, monsieur Pierre, d" etre venu voir une pauvre malade, [Very kind of you, Pierre, that you came to visit the poor patient,] - Anna Pavlovna told him, looking at her in fright with her aunt, which she let him down. Pierre drilled something incomprehensible and continued to look for something with his eyes. He smiled joyfully, cheerfully, bowing to the little princess, as a close acquaintance, and went to his aunt. The fear of Anna Pavlovna was not in vain, because Pierre, having not heard the aunt's speech about the health of her Majesty, departed from her. Anna Pavlovna frightenedly stopped him with the words:
  “You don't know Abbot Morio?” he is a very interesting person ... ”she said.
  - Yes, I heard about his plan of eternal peace, and it is very interesting, but hardly possible ...
  “Do you think? ...” said Anna Pavlovna, in order to say something and to return to her housewife’s occupations again, but Pierre did the opposite. Before he had not heard the words of the interlocutor, he left; now he stopped with his conversation the interlocutor who needed to leave him. He, bending his head and legs apart, began to prove to Anna Pavlovna why he believed that the plan of the abbot was a chimera.
  “We will talk later,” said Anna Pavlovna, smiling.
And, having got rid of a young man who could not live, she returned to her housewife’s activities and continued to listen and take a closer look, ready to give help to the point where the conversation was weakening. As the owner of the spinning workshop, having put the workers in their places, paced around the establishment, noticing the immobility or the unusual, creaking, too loud sound of the spindle, hastily walking, holding back or putting it into the proper course, so Anna Pavlovna, walking around her living room, came up to shut up or speaking too much to the circle and in one word or by moving again brought on a uniform, decent conversational machine. But among these concerns, everything was visible in her was a special fear for Pierre. She carefully glanced at him while he went to listen to what was said near Mortemar, and went to another circle, where the abbot spoke. For Pierre, brought up abroad, this evening Anna Pavlovna was the first that he saw in Russia. He knew that all the intelligentsia of St. Petersburg was gathered here, and his eyes flashed like a child in a toy shop. He was still afraid to miss the smart conversations that he could hear. Looking at the confident and elegant expressions of the faces gathered here, he was still waiting for something especially clever. Finally, he walked over to Morio. The conversation seemed interesting to him, and he stopped, waiting for the opportunity to express his thoughts, as young people like it.

Consider the main advantages of blade systems:

Unique physical design. The architecture of the blade systems is based on a detailed, unique physical design. Sharing resources such as power, cooling, switching, and management reduces complexity and eliminates the problems that are common to more traditional rack-mount server infrastructures. The physical design of blade systems involves placing server blades in a special chassis and its main structural element is the backplane. The backplane is designed in such a way that it solves all the problems of switching blade servers to the outside world: with Ethernet networks, Fiber Channel storage networks, and also provides SAS (SCSI) communication with disk subsystems in the same chassis. The blade chassis also allows you to place the necessary Ethernet or Fiber Channel switches in it to communicate with external networks. Access to these switches from server blades is provided by pre-installed or installed additional controllers. The means of switching to external networks integrated into a common shelf significantly reduce the number of cables for connecting to a LAN and SAN than to traditional rack servers. Server blades have common power and cooling facilities. Placing power and cooling systems in a common shelf, rather than in separate servers, provides reduced power consumption and increased reliability.

Better Management and Flexibility. Blade servers are fundamentally different from rack servers in that the server shelf has intelligence in the form of control modules, which is not present in racks when placing traditional servers. To control the system does not require a keyboard, video and mouse. The blade system is controlled using a centralized control module and a special remote control processor on each blade server. The control system for the chassis and servers usually have sufficiently convenient management software. There are opportunities to remotely control the entire "Blade" -system, including power management and the network of individual nodes.

Scalability - if necessary, increase production capacity, it is enough to purchase additional blades and connect to the chassis. The servers and infrastructure elements in the blade systems are smaller and take up less space than similar rack-mount solutions, which helps save energy and space for IT. In addition, thanks to the modular architecture, they are more convenient in implementation and modernization.

Increased reliability. In traditional rack environments, additional equipment, switching tools, and network components that provide redundancy are installed to increase reliability, which incurs additional costs. Blade systems have built-in backup tools, for example, it is assumed that several power supplies are available, which allows for failure of one power supply to ensure uninterrupted operation of all servers located in the chassis. Also cooling components are duplicated. Failure of one of the fans does not lead to critical consequences. If one server fails, the system administrator simply replaces the blade with a new one and then remotely installs the OS and application software on it.

Lower operating costs. The use of blade architecture leads to a decrease in power consumption and heat generated, as well as to a decrease in the occupied volume. In addition to reducing the footprint in the data center, the economic effect of switching to blades has several other components. Since they include fewer components than conventional rack-mount servers, and they often use low-voltage processor models, which reduces the power and cooling requirements of machines. Blade infrastructure is easier to manage than traditional rack-based IT infrastructures. In some cases, blade systems have allowed companies to increase the amount of resources managed by one administrator (servers, switches and storage systems) more than twice. Management software helps IT organizations save time by efficiently deploying, monitoring, and monitoring blade infrastructure. The transition to a server infrastructure built from blades allows you to implement integrated system management and move away from the previous scheme of Intel servers, when each application was allocated a separate machine. In practice, this means significantly more rational use of server resources, reducing the number of routine procedures (such as connecting cables) that a system administrator must perform, and saving his working time

The advent of storage systems and networks

Another feature of the modern history of the development of computing systems, along with the advent of blade servers, is the emergence of specialized systems and storage networks. Internal server storage subsystems often could no longer provide the necessary level of scalability and performance in the conditions of an avalanche-like increase in the volume of processed information. As a result, external data storage systems appeared, focused solely on solving data storage problems and providing an interface for accessing data for their use.

Data Storage System (SHD) is a software and hardware solution for organizing reliable storage of information resources and providing guaranteed access to them.

Storage systems are reliable storage devices allocated in a separate node. A storage system can connect to servers in many ways. The most productive is the connection via optical channels (Fiber Channel), which makes it possible to access storage systems with speeds of 4-8 Gb / s. Storage systems also have redundancy for the main hardware components - several power supplies, raid controllers, FC adapters and optical patch cords for connecting to FC switches.

Fig. 1.3.

Let's note the main advantages of using storage systems:

High reliability   and fault tolerance - it is implemented by full or partial redundancy of all system components (power supplies, access paths, processor modules, disks, cache, etc.), as well as a powerful monitoring and notification system about possible and existing problems;

High availability   data - it is provided with thought-out functions for maintaining data integrity (using RAID technology, creating full and instant copies of data inside a disk rack, replicating data to a remote storage system, etc.) and the ability to add (update) hardware and software to a continuously working data storage system without stopping the complex;

Powerful controls and controls   - management of the system through a web-based interface or command line, selection of several options for notifying the administrator about problems, full monitoring of the system, performance diagnostics technology working at the hardware level;

High performance   - it is determined by the number of hard drives, the cache size, the processing power of the processor subsystem, the number of internal (for hard drives) and external (for connecting hosts) interfaces, as well as the ability to flexibly configure and configure the system to work with maximum performance;

Seamless scalability   - usually there is the possibility of increasing the number of hard drives, cache size, hardware upgrading of the existing data storage system, expanding the functionality with the help of special software running on the rack, without significant reconfiguration or loss of any storage functionality. This point allows you to significantly save and more flexibly design your storage network.

Today, data storage systems are one of the key elements on which the continuity of the company's business processes depends. In a modern corporate IT infrastructure, storage systems are usually separated from the main computing servers, adapted and configured for various specialized tasks. Storage systems implement many functions, they play an important role in building online backup and recovery systems, fault-tolerant clusters, and highly accessible virtualization farms.

Storage networks

SAN is a high-speed dial-up data network that integrates servers, workstations, disk storage and tape libraries. Data is exchanged using the Fiber Channel protocol, which is optimized for fast guaranteed messaging and allows information to be transmitted over distances from several meters to hundreds of kilometers.

The driving force behind the development of storage networks has been the explosive growth in business information (such as e-mail, databases, and highly loaded file servers) that requires high-speed access to disk devices at the block level. Earlier at the enterprise there were "islands" of high-performance SCSI disk arrays. Each such array was allocated for a specific application and is visible to it as a certain number of “virtual hard drives”. A Storage Area Network or SAN allows these islands to be combined with a high-speed network. The basis of the SAN is a fiber-optic device connection via the Fiber Chanel interface, which provides information transfer speed between objects 1,2,4 or 8 Gbit / sec. Storage networks help increase the efficiency of the use of resources of storage systems, since they make it possible to allocate any resource to any network node. Consider the main benefits of SAN:

• Performance. SAN technologies provide high performance for data storage and transmission.
• Scalability. Storage networks provide the convenience of expanding the storage subsystem, making it easy to use previously purchased devices in conjunction with new storage devices.
• Flexibility. Sharing storage systems typically simplifies administration and adds flexibility since cables and disk arrays do not need to be physically transported and switched from one server to another. SAN allows you to connect new servers and disk arrays to the network without stopping the system.
. Another advantage is the ability to load servers directly from the storage network. With this configuration, you can quickly and easily replace the failed server by reconfiguring the SAN so that the replacement server will boot from the logical drive of the failed server.
• fault tolerance. Storage networks help you recover more efficiently after a failure. A SAN may include a remote site with a secondary storage device. In this case, you can use replication - implemented at the level of array controllers, or using special hardware devices. Demand for such decisions has increased significantly since the events of September 11, 2001 in the United States.
• Control. SAN technologies enable centralized management of the entire storage subsystem.

Currently, there is a large class of tasks requiring a high concentration of computing tools. These may include both complex resource-intensive computing (scientific problems, mathematical modeling, computational search), and serving a large number of users (distributed databases, Internet services and hosting, application servers).

The power of a computing center can be done more by increasing the performance of individual computing modules or their number. Currently, the second trend is prevailing, and the efforts of developers are aimed primarily at the implementation of parallel computing.

This is due to the fact that since the performance of the central processor is very high at a relatively low cost, it is more rational to use cluster configurations for solving most problems, rather than complex multiprocessor systems. Most likely, this tendency will most likely continue (this can be hoped for by the appearance of multi-core central processors, which further reduce the gap in computing power between specialized solutions and simple servers with two or four central processors).

The increase in the number of computing modules in the data center requires new approaches to server hosting. The use of cluster solutions leads to increased costs for premises for data centers, their cooling and maintenance.

To solve some of these problems, a new type of server will help - modular, often called Blade servers, or blade servers. Blade servers that are gaining popularity are truly 21st century servers (their first models were developed in 2001), the advantages of which are described by manufacturers using the “1234” rule. It sounds like this: compared to conventional servers with comparable performance, Blade servers take up half as much space, consume three times less energy and cost four times less.

So what is a blade server? First of all, it is worthwhile to give an IDC definition: an analytics company calls a modular single-board computer system with a processor and memory as a blade (Blade server). The blades are inserted into a special chassis (or shelf) with a backplane, which provides them with a network connection and power supply. This blade chassis, according to IDC, is a Blade system. It is designed to be installed in a standard 19-inch rack, and depending on the model and manufacturer, it occupies 3U, 6U or 10U in it (one U-unit, or mounting unit, is 1.75 inches). Due to the common use of components such as power supplies, network cards and hard drives, Blade servers provide a higher density of rack computing power compared to conventional thin servers 1U and 2U in height, such as dell pe r210 or dell pe r410.

In fact, the blade system consists of the following components

• Blade servers (in fact, these are ordinary servers without a power supply, with passive heatsinks and without PCI connectors - all “unnecessary components” have been removed)
• Enclosure and passive Backplane (board for switching installed equipment)
• Power and cooling systems (fans, power supplies)
• External switching devices (Ethernet, FC, Infiniband)

Instead of the usual PCI (PCI-E, PCI-X) cards, mezzanine cards are inserted into the server, which allow the use of FC, Infiniband, SAS, or additional Ethernet ports, if the chassis has an appropriate external switching module.

Typical Blade server: no power supply, no fans - nothing more!

Nevertheless, according to analysts, the increased density of blades is now fading into the background and their main advantage for the corporate sector is improving server manageability with a higher degree of automation of their maintenance. The transition to a server infrastructure built from blades allows you to implement integrated system management and move away from the previous scheme of Intel servers, when each application was allocated a separate machine. In practice, this means significantly more rational use of server resources, reducing the number of routine procedures (such as connecting cables) that a system administrator must perform, and saving his working time.

Typical 10U chassis for 10 blade servers

In addition, Blade servers are much easier to maintain than regular rack servers - for example, when a machine crashes, the system administrator simply replaces the blade with a new one and then remotely installs the OS and application software on it. Currently, developers of packages for managing Blade servers implement in their products not only the possibility of automatic deployment on new OS servers and applications, but also the functions of quickly updating installed software or installing software “patches”.

Blade servers are an extremely effective solution for saving space in data centers (DPC), as well as in terms of their consolidation and transition to centralized management of the server park. For example, a system administrator can manage a chassis with blades as a single object and, as the load grows, increase its computing power by adding new blades. In addition, since it is usually possible to install network switches in the chassis, this option also allows consolidation of network resources of the data center.

In addition to reducing the footprint in the data center, the economic effect of switching to blades has several other components. Since they include fewer components than conventional rack-mount servers, and they often use low-voltage processor models, the energy and cooling requirements of machines are reduced. As mentioned above, the administrator’s working time is saved, which, as a result, manages to service more facilities, and therefore, with the growth of the server park, the enterprise does not have to hire another administrator. Finally, although the chassis needs to be purchased along with the blades when moving to the Blade server architecture, thanks to the sharing of its components, further scaling of the Blade system is less expensive than rack-mounted servers, and a chassis with five to ten blades is cheaper than a similar number conventional rackmount machines.

The advantages of using Blade servers can be expressed in the following list:

• cost reduction and increased reliability of the power and cooling system;
• reduction in the number of patch wires;
• improving the convenience of managing the system;
• decrease in occupied volume;
• reduction in energy consumption and heat generated;
• high scalability;
• flexibility.

Of course, “Blade” also has disadvantages:

• lack of a common standard and common platform. For users, this translates into the inability to install, for example, a SUN Blade server in an IBM chassis;
• one point of failure in most Blade systems: the chassis itself is the point of failure of the system. If the chassis fails, then all the “blades” installed in it are idle, which undoubtedly causes significant damage to the business of their owner. However, it must be understood that the chassis itself (body and backplane) is a passive component that can only be physically damaged;
• “Blade” are ineffective for solving tasks requiring a small number of servers (for example, two or three). In this case, it is more profitable to use ordinary servers without overpaying for the chassis;
• a limited class of applications executed by Blades: since Blade servers are equipped with one or two processors, tasks optimized for traditional servers with a large number of processors are difficult to transfer to Blade;
• poor internal scalability of Blade-servers: lack of the ability to install expansion cards in servers (the ability to install one low-profile PCI-card in some models), small amounts of RAM.
• limited size of the internal disk subsystem - the server usually has a connector for 2 2.5 hard drives. This is due to the fact that blade servers are used for virtualization tasks and the use of an external array is implied.

In general, the introduction of Blade-servers brings more positive effect, it is cost-effective. However, traditionally, people resist the introduction of innovation. Most companies are wary of acquiring Blade systems, preferring time-tested rack or tower servers. But every year the sales volume of Blade-servers is growing, there are more and more solutions based on them.

How are Blade servers different from conventional rack servers?

Physical construction.   Many of the advantages of Blade systems provide a unique physical design. Sharing resources such as power, cooling, switching, and control reduces complexity and eliminates the problems that are common to more traditional server infrastructures. The physical design of Blade systems involves placing Blade servers in a special chassis (shelf) and its main structural element is the backplane.

The backplane is designed in such a way that it solves all the tasks of switching Blade servers with the outside world: powering up, connecting to Ethernet networks, Fiber Channel networks and also provides SAS (SCSI) interaction with storage systems in the same chassis (additional drives or tape). The chassis for Blades also allows you to place the necessary Ethernet or Fiber Channel switches in it for communication with external networks. Access to these switches from the Blade servers is provided by pre-installed or optional Ethernet or Fiber Channel controllers.

Fewer cables.   The means of switching to external networks integrated into a common shelf significantly reduce the number of cables. The new HP BladeSystem requires 94% less LAN and SAN cabling than traditional rack servers. Switches can be inserted into the chassis, which provide interaction both inside the chassis and with external devices.

If you think that there are a lot of cables - you are mistaken

General power and cooling facilities. Placing power and cooling systems in a common shelf, rather than in separate servers, provides reduced power consumption and increased reliability. In addition, thanks to new and more intelligent features, energy consumption is reduced. For example, for HP Blade servers, the new HP Thermal Logic cooling technology reduces overall energy consumption by more than 30% compared to rack-mount systems, and requires less energy and air to cool the entire solution. Basically, this is achieved by using more powerful power supplies and fans common to the entire chassis.

The best management features.   Blade servers are fundamentally different from rack servers such as dell pe r510 in that the server shelf has intelligence in the form of control modules. For HP Blade systems, the intelligent management functions are performed by the dedicated Onboard Administrator module, which is not available in racks when hosting traditional servers. In addition to IP KVM, there are tools for monitoring and power management, switching modules and tools for rapid deployment of the system.

Easy administration of any blade server in the chassis

No keyboard, video, or mouse required.   The new HP BladeSystem is managed through a centralized Onboard Administrator module and an iLO2 remote control processor on each Blade server, so no separate cables and switches for keyboard, video and mouse (KVM) are required.

Size and density.   The servers and infrastructure elements in the HP BladeSystem are smaller and take up less space than similar rack-mounted solutions, which helps save energy and space for IT. In addition, thanks to the modular architecture, they are more convenient to implement and maintain.

The main advantage of Blade servers is increased density.

Increased Reliability.   In traditional rack environments, additional equipment, switching tools, and network components that provide redundancy are installed to increase reliability, which incurs additional costs. For example, the HP BladeSystem has built-in redundancy, with N + 1 redundancy configurations more convenient to configure and cost-effective than full redundancy.

Lower operating costs. Blade systems infrastructure is easier to manage than traditional rack-based IT infrastructures. In some cases, HP BladeSystem systems have allowed companies to increase the number of resources managed by one administrator (servers, switches and storage systems) more than double. Based on the average, the deployment time for one ProLiant DL360 rack server is 4 hours, and the ProLiant BL460c Blade server is 30 minutes. After installing the server shelf, adding a new Blade server is much faster than installing a rack server, which includes installing the server in a rack, laying cables, installing the operating system, and setting up connections to the LAN and storage systems. Any changes are also faster. HP Insight Control management software helps IT organizations save time by efficiently deploying, monitoring, and monitoring the HP BladeSystem infrastructure. By automating core IT processes, it allows employees to focus on more important tasks and act proactively.

Blade Server Evolution

The first company to start manufacturing Blade servers was RLX Technologies. Since RLX is the ancestor of Blade servers, their history is inextricably linked with the development of the company's lineup. Let us dwell in more detail on the development history of Blade servers manufactured by RLX Technologies.

First Blade Server RLX ServerBlade 633

In May 2001, the first-generation Blade server, ServerBlade 633, appeared. To reduce the amount of heat generated, RLX engineers decided to equip the Blade with low-voltage Transmeta Crusoe processors used in laptops. Unfortunately, this solution turned out to be ineffective: in comparison with ordinary rack-mount servers, Blade lost - were less powerful and overheated.

Second generation servers, Serverblade 667, which appeared already in December 2001, had a performance increase of more than 30% compared to ServerBlade 633. In addition, it was possible to install “blades” of the first and second generations in the same chassis, which simplified the transition to more fast "blades". The second-generation blade servers, although they, like their predecessors, were equipped with low-voltage Transmeta Crusoe processors, surpassed them in memory size and could work with a large number of operating systems: various configurations of Windows Server 2000 and Red Hat Linux. Another innovation in RLX’s second-generation Blades is the separate Blade Management Network and Blade Management System, known as the RLX Control Tower Blade. The latter monitored the “blades” and managed them as a single system, and also provided a remote console to the “blades” over the network. In addition, the RLX Control Tower Blade control system had a Web-based interface, which allowed administration of the entire system at a lower cost. RLX’s second-generation “blades” were used to build Linux clusters in areas such as bioinformatics, seismic analysis, and computational chemistry.

In third-generation servers, released in February 2002, the problem of overheating processors was resolved: a low-voltage 800-MHz Intel Pentium III processor was installed on the ServerBlade 800i. With the basic configuration with 512 MB of memory, it was possible to expand the memory to 1 GB, and the total amount of hard drives left on the “blades” could be increased from 20 to 80 GB. This generation, in addition to increasing power, was distinguished from the previous one by the appearance of a “blade” of a new type - the manager. It played the role of an integrated management bus, and with it, the system administrator could configure the switch and VLAN ports from the command line. Like second generation servers, Serverblade 800i   has been optimized for various operating systems of the Windows and Linux families.

One of the first Blade servers

Fourth Generation Blades - Serverblade 1200ithat appeared in September 2002 were 50% faster than their predecessors in terms of processor speed (they were equipped with Intel Pentium III CPUs with a frequency of 1.2 GHz and 512 KB cache in the second level), the capacity of the disk subsystem (it has not yet been taken out of Blades ”, the maximum total capacity of local drives was 120 GB) and 100% in memory (maximum - 2 GB).

In fifth generation servers - Serverblade 2800i, 3000i, released in March 2003, it is interesting not only to increase processor performance, memory and disk subsystems. These were the first multiprocessor “blades” - they were equipped with two 2.8-GHz Intel Xeon processors with 512-KB L2 cache. In addition, RLX Technologies was the first company to equip the manufactured “blades” with Infiniband interface cards. Apparently, these two enhancements significantly increased the price: ServerBlade 2800i was twice as expensive as ServerBlade 1200i.

The last, sixth generation Blade RLX SB6400, was released in November 2004. The maximum memory capacity was increased to 4 GB, they supported up to two 64-bit Intel Xeon processors with a clock frequency of 2.8 to 3.6 GHz. They integrated Ethernet, Infiniband and Fiber Channel network interfaces, as well as a PCI Express controller.

However, despite the good characteristics of the new servers, the pioneer of the Blade architecture had to leave this market. RLX acknowledged that competing in the market with giants such as Sun, HP and Dell is too difficult. She decided to curtail the production of “blades”, limiting her activities to the production of managing software.

On October 3, 2005, RLX Technologies was acquired by HP, which continues to produce RLX “blades,” and RLX has become the HP technology group.

By 2005, all major players in the server hardware market became interested in Blade servers and began to release their own hardware. Some companies developed their solutions in the field of Blade-systems, others borrowed existing solutions. Fierce competition was created for the Blade-systems market. Each vendor tried to bring his innovation to the “blades”.

HP ProLiant 460c state-of-the-art Blade Server: more compact and more economical

Currently, for example, one HP ProLiant 460c Blade server is not inferior in performance and fault tolerance to the thin HP ProLiant DL360 rack-mount server, 1U high. In this case, in 1 Blade-chassis HP BladeSystem, a height of 10U can accommodate 16 servers HP ProLiant 460c. Thus, in a standard rack, 42U high, 42 HP ProLiant DL360 G6 servers or 4 x 16 \u003d 64 HP ProLiant 460c servers will fit. At the same time, Blade servers are easier to manage and administer; they will consume less power while generating less heat. The work of maintenance personnel is also reduced due to a decrease in the number of cables suitable for the server rack by several times.

Gradually, the Blade servers were able to use most of the technologies used in conventional rack servers. Full-featured RAID controllers using hot-swappable drives have appeared. The Blades support popular communication technologies — Gigabit Ethernet network adapters and switches, fiber optic adapters and Fiber Channel switches, and high-speed Infiniband connections. Now in most blade systems there is no single point of failure - all the main components of the system are duplicated for fault tolerance. And if the Blade server fails, it can be replaced entirely and restored to work in a short time.

Instead of a Blade server, a Blade drive can be installed in the chassis

The latest innovations in the world of Blade servers include Storage-Blade (storage), Tape-Blade (tape) and PCI-Blade. The first are blades that are clogged with hard disks as much as possible, and in fact are external disk baskets for ordinary “blades” installed in the chassis. The second are blades, in the form factor of which a modern tape drive fits. The third of these “blades” are blades for installing full-size boards with PCI, PCI-x, or PCI-e interfaces. They are designed to solve the problem of internal scalability of Blade servers.

Thus, in an effort to consolidate resources, the manufacturer offers companies a single unit that combines computing resources (Blade servers) with expansion options that are not inferior to conventional servers (support for multiple multi-core processors and large amounts of memory “blades”, internal mezzanine boards plus “PCI Blades”), storage systems (“Storage Blades”) and backup devices (“tape blades”).

Blade Server Evolution

The Future of the Blade Server Market

Currently, the Blade server market is booming - new blade models are coming out, both hardware and software are being updated. However, the following fact prevents us from building optimistic forecasts. Despite the fact that, on the one hand, the solutions based on the Blade-North are not inferior in many respects to classical solutions (for example, in terms of power, scalability, manageability, reliability) and are often unique, on the other hand, it is the architectural features that lead to that they often look at Blade servers with caution as a violator of established traditions. Another reason is the relatively high cost of implementing systems. If the expansion of the company's IT system is usually carried out by installing additional computers, then a chassis even with a small number of installed blades can cost more than individual servers of similar capacity, especially if the solution involves the use of a non-trivial data storage system. Perhaps the “virtual” fear of being tied to one supplier also plays a role, since the system requires significant investments and is not always easily integrated into the existing software and hardware park. Clear planning for the development of the company's IT structure and choosing the right supplier can help with these problems.

A negative role in the development of the Blade solutions market is played by the lack of open standards for chassis hardware design (and this despite the fact that often solutions from different companies are, in fact, small variations of one design). The situation with the software is a bit simpler, since modern blades are based on standard equipment and, if desired, you can implement everything using popular software.

With the advent of Blade systems, the terms consolidation and virtualization of the information infrastructure have become increasingly mentioned. Heads of IT departments have long been looking for a way to eradicate patchwork infrastructure by combining all key infrastructure elements into one single system controlled from a single management console. Already now it is possible to combine the Blade-servers of the chassis or the whole rack into a “single, large” virtual server and allocation of the exact amount of production capacity to solve a particular problem. Blade-servers simplify the creation and use of virtual networks (VLANs), as well as bring extraordinary ease in their management.

HP experts predict that in the near future tower servers will leave the market, only rack-mount servers will remain, including the already mentioned representatives of the 6th generation, HP ProLiant DL360G6, HP ProLiant DL380 G6, HP ProLiant DL180 G6 and Blade servers. Moreover, the latter will be distributed more, and will gradually push rack servers out of the market. According to analysts, by 2009, the share of blades from the total server market will exceed the threshold of 30%, and sales will reach $ 9 billion (data from the IDC analytical service).

Samples of Modern Blade Servers

Let's compare three popular Blade systems.

• Side indicators are within normal limits.
• Input factors are the same for all considered technical systems.
• Prices are approximate, accurate enough to compare systems with each other.
• Performance was evaluated in points based on the maximum number of computing processors and the amount of RAM supported by the system.

HP BladeSystem C-class

IBM BladeCenter H

HP BladeSystem c-class   and IBM BladeCenter H   They are the closest competitors and have similar technical performance. These are the most prominent representatives of the Blade systems market. We call them the highest class Blade systems.

Top-class Blade server subsystems: processor, system bus, RAM, interface controllers, disk controller, hard drives, backplane, cooling modules, power modules, separate control module, module for local connection to the Blade system, interface switches, module for connecting PCI cards.

Dell PowerEdge 1955

Dell PowerEdge 1955   is a simpler system, not burdened with additional "twists", but, nevertheless, showing good performance and does not take up much space in the rack. The Dell PowerEdge 1955 is a mid-range Blade system.

The number of technical parts of mid-range Blade systems: processor, system bus, RAM, interface controllers, disk controller, hard drives, backplane, cooling modules, power modules, separate control module, interface switches.

Analyzing the results, we can conclude that the Blade-systems of the highest and middle level have not large coefficients of imperfection. The structure of the Blade systems is close to optimal. Although it consists of a large number of technical parts, each part in it plays its own unique role, and they all create what they call the Blade system.

Comparing the results obtained in the course of determining the perfection of technical systems with the data on sales volumes in recent years, we can say that, despite the greater perfection of middle-class Blade systems, top-class Blade systems are better sold. This is due to the fact that such giants of the IT industry as HP   and Ibm. The prestige of the brand, successful promotions, as well as good conditions for providing technical support from the manufacturer, encourage the buyer to purchase such systems. Blade systems of the highest level are also equipped with the largest number of technological innovations, which is very attractive to the buyer.

When should I switch to blade servers?

At the time of writing the review in the Russian market, HP supplied Blade servers to several hundred customers. They belonged to various vertical markets (banks and finance, industry, trade, distribution of goods, science and education, transport, government agencies and departments, etc.). Accordingly, both the size of customers and the number of servers in their processing centers also differed significantly.

In its most general form, the main motive for the transition to the purchase of Blade systems is the general modernization of the server farm and the desire of customers to move from the existing server zoo to a more modern and streamlined server system in the organization. In these cases, the customer stops at choosing Blade systems as the most modern and promising solution for standardizing servers in the organization.

An important additional argument in favor of choosing Blade systems is that the customer has a real or planned solution for consolidating the storage system. In favor of this consideration, one can also cite the fact that requests for Blade server specifications very often involve ordering Blade systems for HP EVA or MSA datasets. Indeed, the simultaneous transition to blades and consolidated data storage has the advantage that it is no longer necessary to store large amounts of data in the external memory of the Blade servers themselves, and in all other parameters, modern blades are not inferior to their rack-mount counterparts.

Turning to more specific cases of introducing new server technologies in organizations, it should be noted first of all that Blade servers have become the preferred and most frequently requested solution for virtualizing and consolidating server applications. In our experience, more server application consolidation projects are being implemented on Blade servers and shared storage systems than on rack servers. Logical explanations can also be found for this fact. Blade servers today are offered not only for the most common two-socket servers, but also for four-socket ones, both on Intel and AMD processors, as well as if we are talking about HP servers and Intel Itanium processors. These servers have very large capabilities in terms of main memory (up to 64 and 128 GB) and can be used to consolidate a large number of very capacious server applications. At the same time, for server blades, server replacements and application transfer are more dynamic and convenient, which determines the preferences for Blade systems.

The choice of blades for implementation on servers of various terminal systems is also very common. And here are the benefits and convenience of the server construct, the lack of a large number of connecting cables for shelves with Blade systems, convenient visualization of the shelf and the quick identification of individual modules in it are arguments in favor of terminal systems on Blade servers.

It is necessary to note one more more specific case when the choice of Blade systems dominates when ordering servers. This is an increasingly widespread computing cluster based on servers of standard architecture. The number of such solutions is constantly growing and their areas of application are actively expanding. Here, the choice of blades, especially for large clusters, certainly looks justified in terms of space requirements, rack space, cooling system requirements, and for this reason the vast majority of computing clusters are ordered on Blade systems.

We thank ServerUnit (Blade servers of any complexity)www.server-unit.ru   for help in preparing the material.

Vladislav Sharov

According to most experts, blade servers (blades, modular, ultra-thin servers) are the next-generation computer architecture, which removes the acuteness of many existing problems and affects the approach of equipment manufacturers to designing standard devices. Such servers are really capable of increasing the density of equipment in the racks, and in addition, they are characterized by better manageability, lower power consumption, more convenient to maintain, easier to deploy, which ultimately leads to a reduction in the total cost of ownership (TCO). And although the first incarnations of server modules best meet the expansion objectives towards the external (client) part of the IT infrastructure, most likely this architecture will continue to develop and will play an important role at all levels of the data center infrastructure.

Essentially, a blade server is a computer assembled on one board. Such a single backplane actually contains the entire computer system, including processors, memory, network connections and other electronic circuits. Some server modules may have a number of local disk drives with the ability to access external storage media.

Each server module runs under its own copy of the OS, therefore, administrators can distribute specific modules between different tasks or users, as well as insert or remove them in hot mode, without interfering with the work of neighboring modules. Blade servers are inserted into the chassis or chassis backplane, which provides shared and redundant resources such as power supplies, fans, network equipment, and cables.

There are several reasons for the success of blade servers. So, systems of this class allow the supplier of computer equipment to provide customers with solutions that increase the efficiency of the data center and reduce the total cost of ownership. Generally speaking, improving the efficiency of the data center is based on several features of blade servers.

Blade servers are relatively cheap because, unlike traditional ones, they do not use individual chassis and their own infrastructure. Sharing power, cooling, control, and cabling systems across multiple servers can significantly reduce costs per server. Now, server hardware vendors can integrate components that users previously had to purchase separately into their designs - this applies to remote control cards, Ethernet switches, and even KVM adapters (Keyboard, Video, Mouse - keyboard, video card, and mouse), which provides a significant reduction hardware cost. Sharing high-quality components between several servers is justified from the point of view of reliability, since this approach reduces the number of components and, accordingly, the probability of their failure. Due to the high initial costs of developing the chassis, blade servers are most suitable for use in a data center environment with a high concentration of equipment per unit area, and not for creating a system of several data centers remote from each other with a small number of servers in each.

Deploying multiple servers can be a very time-consuming task, requiring large resources. The administrator installs each server in a rack, connects network and power cables and installs software on them. In an environment with a high density of equipment, one of the main problems that are often encountered during installation is associated with the cabling. Using blade servers, the administrator installs and connects only the chassis, since individual modular equipment, as a rule, does not require cable connections. Adding computing resources in this case comes down to simply adding a new board. Having a common chassis minimizes the problem of cabling, since redundant power supplies simultaneously serve multiple modules. In addition, on-board switches consolidate Ethernet and even KVM management.

Automated software installation tools allow administrators over the network to quickly and easily deploy software on one or many server modules simultaneously. After installing the software on the blade servers, the administrator can fully manage them using remote administration tools. In this case, customers should rely on manufacturers to supply a complete set of tools for remote software installation, hardware configuration, and system management. This toolkit enables customers to manage blade and traditional integrated servers from a single structure, which reduces costs.

As already noted, the maintenance of a failed module is reduced to replacing it with a working one. An administrator can also easily create a global hot-spare backup server that, through remote control, can perceive all the attributes and tasks of a blade server that needs maintenance or repair.

Since the premises in which data centers are located are very expensive, in the last couple of years there has been a tendency in the market to use server racks optimized in density. Blade servers in comparison with today's optimized rack-mounted server designs 1U high (1U \u003d 1.75 inches \u003d 4.445 cm) can increase the density of equipment several times, depending on the manufacturer and the design used. In addition, since each module is an independent server, the chassis allows you to combine the execution of several applications within a single enclosure. Such consolidation also helps to achieve a higher degree of centralization of computer resources.

Heads of data centers should assess the needs of each client and make sure that they allocate such volumes of shared infrastructure resources (whether it’s their own unit or an external customer in the case of service providers) that would ensure work under conditions of peak capacity utilization. It often happens that a significant amount of equipment remains underutilized, while consuming energy, generating heat and spending money, administrative resources, power and space.

The transition to the architecture of blade servers and the use of related improvements in load management software allows us to calculate the size of the infrastructure according to peak load, not for individual clients or groups, but according to the needs of the entire data center as a whole. This task is usually accomplished with management software, but blade servers can improve software performance due to their low cost and simplified process of adding computing resources.

To reduce power consumption, most blade server manufacturers try to use low power processors in some of their designs. But even with conventional processors, blade servers will consume less power than regular servers because they use fewer components requiring high power. In addition, manufacturers can use more efficient power supplies, since these units serve many servers. Ultra-compact servers that consume less energy, respectively, emit less heat, which, in turn, reduces the amount of energy required to cool systems.

Fujitsu Siemens Computers Blade Servers

Fujitsu Siemens Computers Corporation (http://www.fujitsu-siemens.com) released its first blade server, the Primergy BX300, in mid-2002 based on the low-voltage version of the Pentium III processor. Each Primergy BX300 server chassis accommodates up to 20 3U server modules that now use the new Pentium M processor clocked at 1.4-1.6 GHz. System memory type DDR SDRAM up to 4 GB supports ECC error detection and correction technology. Two IDE hard drives with a capacity of up to 60 GB each can be combined into RAID 0, RAID 1 and RAID 7 (JBOD) arrays. There are two Gigabit Ethernet channels for each server module.

Thanks to all this, the Primergy BX300 is able to provide the operation of complex and resource-demanding systems, for example, a large number of terminal servers or Web servers. At the same time, such an advantage as reducing costs per one blade server and reducing the required space in the rack is provided.

The new Primergy BX300 server module also includes an optional PCI slot for connecting to Gigabit Ethernet networks and Fiber Channel. This means that the BX300 can provide even greater bandwidth by integrating into corporate networks or providing additional features in network storage infrastructures. When deploying blade servers, this aspect is often vital for Internet service providers and applications.

The BX300 chassis includes redundant and hot-plug power modules and fans, as well as an integrated KVM switch to provide direct access to each individual server module. Autonomously functioning blade servers equipped with a new processor with energy-saving technology increase productivity in data processing and computing, and a stable, high-performance connection of Primergy blade servers to a local area network is guaranteed by redundant switch modules that are connected to Gigabit Ethernet networks in a hot mode. At the same time, the control modules, also duplicated and connected in hot mode, provide remote administration via a local network or connection via a dedicated serial port - regardless of the state of the system. In addition, there is a solution for installing software that has a lot of opportunities for simple and efficient distribution of programs among individual modules and cloning of already installed packages. Status and location information is indicated by LEDs on both the chassis and Primergy server modules.

In early 2004, the corporation added its Primergy BX600 high-performance system to its blade server family. The configuration items also included power modules, switch modules, and control modules. Each case (shelf) of the Primergy BX600 contained up to 10 server modules, including an Intel Xeon processor with a clock frequency of 2.8 GHz.

The Primergy BX600 blade servers were designed primarily for the mid-tier application industry and therefore can serve as a platform for virtually all data center applications. They perform the functions of both high-performance application servers and small and medium-sized database servers due to the large amount of RAM and a variety of connectors that provide connection to a local area network using the Gigabit Ethernet standard, storage systems via 2 Gb / s Fiber Channel ports and backup devices according to the SCSI standard.

The Primergy BX600 provides not only classic mid-tier applications, such as Microsoft Exchange or Lotus Notes, but also enterprise-class applications, including FlexFrame for mySAP Business Suite, traditional SAP applications, Siebel, Baan, I2 and the like. The Primergy BX600 serves as a good platform for databases such as Oracle 9iRAC and especially the Oracle 10g version that supports scaling. All other Primergy BX600 components (for example, integrated Ethernet switches) provide fairly high performance and bandwidth.

At the end of the summer of 2004, Fujitsu Siemens announced the start of sales of the four-processor Primergy BX600 models, suitable for computer centers with high demands on computing power. In addition, the Primergy BX600 meets the requirements of modern corporate computer centers, Internet service providers and application rental service providers. The ability to install up to 30 four-processor servers in a 19-inch rack with a height of 46U ensures optimal use of the area of \u200b\u200bthe premises of computer centers. Blade servers include additional components that provide reliability and fault tolerance, as well as convenience and management efficiency. The bundled Primergy Server Management Suite helps you avoid problems installing and configuring systems.

The module can now be installed up to four Intel Xeon MP processors, up to 16 GB of RAM, as well as four Gigabit Ethernet channels and up to four Fiber Channel channels for connecting SAN systems. Each server module includes two Ultra320 SCSI hard drives with hot-swappable capacity of 146 GB each. For their installation does not require a separate modular connector.

Since the four-processor modules use the same Primergy BX600 infrastructure as the two-processor modules, they can be installed in one housing (mixed configurations). The Primergy BX600 chassis can contain up to five four-processor servers or up to ten two-processor servers.

HP Blade Servers

As early as the spring of 2002, HP Corporation (http://www.hp.com) introduced a unified family of ultra-compact servers that included the HP Server BH and HP ProLiant BL single-board server lines. According to experts, with two lines of blade servers, HP was able to demonstrate the industry’s most comprehensive and powerful portfolio of products, technologies, administration software, tools, services and partnerships in this area.

The first model in the ProLiant BL e-Class series was the BL10e server, released by Compaq (later merged with HP) in early 2002. Up to 280 of these servers could be installed in a 42U rack (14 shelves with 20 servers each). Currently, the ProLiant BL e-Class (BL10e G2 model) uses a Pentium M 1 GHz ULV processor with ultra-low operating voltage. The server is equipped with SDRAM RAM from 512 MB to 1 GB, a 40 GB hard drive and two 10/100 Ethernet ports.

The adaptive infrastructure strategy developed by HP aims to enable companies to quickly and efficiently respond to changes and optimize their client infrastructures with a view to the future. This opportunity is clearly demonstrated by new hardware and software technologies, in particular, the architecture of the ProLiant BL blade servers.

Today, the HP ProLiant BL line is the industry’s most comprehensive set of blade servers and technologies optimized for the challenges faced by data centers, service providers and telecommunications operators. HP blade servers reduce the cost and complexity of information systems through comprehensive management and more complete implementation of server and storage capabilities. The corporation offers a wide selection of blade servers that provide front, intermediate and rear software tools and systems.

Thus, the HP ProLiant BL10e G2 blade servers are designed to work on the edge of the network and infrastructure applications, for example, as Web servers, firewalls, proxy or caching servers. This model still occupies a leading position among servers of this class in terms of energy efficiency. You can install 20 ProLiant BL10e server blades in a 3U chassis, and 280 servers in a 42U full-size rack. Thus, optimized use of the expensive space of computer centers, as well as power consumption and cooling equipment with a minimum number of cables.

The HP ProLiant BL20p G2 server blades are designed for use as arrays of terminal servers, Web hosting systems, e-commerce, and streaming media services. At the same time, they are able to protect important information stored on servers. Each module supports up to two Xeon MP processors with a clock frequency of 2.8-3.2 GHz. The amount of RAM is from 512 MB to 8 GB; the capacity of internal SCSI drives is 292 GB.

Of particular note is the ProLiant BL30p system, which, according to the developers, has no analogues among the blade servers of other manufacturers, for example, by the elegance of solving the problem of increasing the density of computing power. The fact is that 96 dual-processor computers are placed in one rack - this is a record figure for blade servers based on Xeon processors. The modules use processors with clock frequencies from 3.06 to 3.2 GHz. Like the BL10e, this dual-processor server has a height of 3U, but is placed in pairs in a 6U-sized shelf. The system is well suited for mid-tier applications that require high processor power. As an option, it provides for the installation of two internal 2.5-inch ATA hard drives (instead of taking up more space, although more productive SCSI-drives in the ProLiant BL20p). The capacity of internal drives is 120 GB, and the amount of RAM reaches 8 GB. The system is oriented to work with SAN and supports booting from external drives, therefore it can be used in a diskless configuration.

It is worth noting that HP was the first among the major server manufacturers to introduce four-processor blade servers. This event was greatly facilitated by the leading position of the corporation in the global server market based on Intel processors. The HP ProLiant BL40p Server has expanded its range of blade servers, which includes single-, dual- and quad-processor models.

The HP ProLiant BL40p server is designed for systems that require multi-processor support and SAN connectivity — such as large enterprise arrays of application servers, messaging systems, CRM and ERP, which also require a reliable, high-availability platform. This server includes up to four Xeon MP processors 2.0 GHz (or 3.0 GHz), PC2100 ECC SDRAM (maximum 12 GB), Integrated Smart Array 5i Plus controller and four hot-swap SCSI disks (total capacity 587 GB ) Using ProLiant BL40p with other servers in the BL family reduces costs by remotely controlling from anywhere in the world with the enhanced Integrated Lights-Out (iLO). In addition, an integrated management system and full compatibility with Insight Manager Suite provide maximum virtual presence, allow you to monitor, administer and monitor the status of the system with the visualization of all applications at the level of each blade server. The duplicated Fiber Channel interface takes full advantage of the external storage of data on the network. By introducing SAN, customers can improve data availability, easily scale capacity and reduce management costs by consolidating disk resources. In addition, HP-led SAN and NAS convergence helps build a storage architecture that includes applications, databases, and file server features. HP recommends the use of HP StorageWorks-branded SANs and NAS NAS, which are specifically optimized for use with company blade servers.

HP ProLiant BL40p Servers.

If it is possible to connect to the SAN, corporate applications can run on blade servers, which will help reduce the cost of managing them, as well as realize the benefits of an integrated system that includes servers, a network, and data storage facilities. In addition, the ability to connect to the SAN allows you to increase the storage capacity of up to hundreds of terabytes; expand support for enterprise computing systems; Connect blade servers and databases to both SAN and NAS / SAN combined systems (for storing both file and block data) realizes the benefits of managing networked storage with the HP OpenView Storage Area Manager software suite.

ProLiant BL p- and e-class servers are backward compatible with existing BL enclosures to help protect customer investment. All HP ProLiant BL family server blades are easily managed with HP ProLiant Essentials, HP Insight Manager 7, and HP OpenView software. The HP ProLiant Essentials Rapid Deployment Pack enables you to automatically configure and install OS and applications on hundreds of servers at once. These technologies are specifically optimized for integration with the HP OpenView software suite, which ensures seamless management of IT infrastructures, voice and data networks.

IBM Blade Servers

IBM (http://www.ibm.com) entered the blade server market relatively late, announcing its first BladeCenter model in the fall of 2002. Nevertheless, according to analysts, it is now, as usual, among the leaders.

As already noted, the advantages of blade servers compared to conventional thin servers are not limited only to a higher density and cost reduction (due to the reduction in the number of cables and other components of the server complex). Perhaps the biggest gain from the use of blade servers is associated with a reduction in the total cost of ownership due to the high flexibility and scalability of these systems, their better manageability and serviceability.

IBM, in particular, intends its BladeCenter system for servicing e-mail servers and organizing collaboration between Microsoft Exchange and Lotus Domino, Citrix MetaFrame terminal servers, Web applications, as well as databases and front-end components of ERP systems. In addition, IBM blade servers are suitable for building high-performance Linux clusters and serving DNS servers, caching, and load balancing.

Currently, the BladeCenter line includes three server modules, which are made in the 7U construct and can be installed in one shelf (chassis). This is based on Intel Xeon (with support for EM64T) dual-processor HS20 and four-processor HS40, as well as using a 64-bit PowerPC 970 processor dual-processor JS20.

It is worth noting that Intel (http://www.intel.com) considered the BladeCenter design the most promising for blade servers, and at one of its Developer Forums (IDFs), the corporation presented a series of construction materials designed for OEM partners developed jointly with IBM Intel Enterprise Server Blade Today it includes a dual-processor computing module Intel Server Compute Blade SBLX52 and a four-processor SBX44 (analogs of IBM BladeCenter HS20 and HS40, respectively).

Despite their common roots, the IBM and Intel blade servers have a number of differences. So, if the proprietary IBM Director proprietary software package for IBM eServer xSeries servers is used to manage BladeCenter, then Intel Server Management performs the same functions in Intel Server Blade, and OpForce developed by Veritas Software is used to automate the deployment of large blade server configurations.

In addition to server modules, you can install expansion modules in the BladeCenter chassis that allow you to use Myrinet and Fiber Channel interfaces, increase the throughput of the Ethernet subsystem, connect additional SCSI and ATA 100 drives. In addition, a four-port Gigabit Ethernet switch is supplied as options for the BladeCenter chassis, Nortel Networks-developed Layer 2-7 Gigabit Ethernet switch, dual-port Fiber Channel switch, Optical Pass-thru module, second control / KVM module, 1800 W power supply and dedicated noise reduction module.

IBM also offers PowerExecutive energy management software. This is the first product as part of a strategy to reduce energy costs in data centers. PowerExecutive is part of the BladeCenter platform management module and allows IT services to better manage the power consumption of blade servers.

Kraftway Blade Servers

Kraftway (http://www.kraftway.ru), one of the recognized leaders in the Russian computer equipment market, first introduced a new class of servers based on Intel Xeon processors under the GEG Express Blade trademark in September 2003. These servers were created on the basis of The 7U Intel Enterprise Server Blade enclosures accommodate up to 14 compute modules (SBXL52) that support one or two Xeon processors each. Note that the Intel Enterprise Server Blade format allows you to create powerful and flexible highly integrated computing solutions. Each computing module supported from 4 to 8 GB of DDR200 standard RAM with ECC error correction code, occupied one housing slot and had two integrated Gigabit Ethernet channels for connecting to the network. Additionally, it was possible to install a dual-channel 2 Gb / s Fiber Channel module, which would provide server integration into existing SAN infrastructures.

Kraftway GEG Express Blade Servers.

In February 2004, Kraftway, being at that time still the only Russian manufacturer of blade servers, announced a four-processor module (SBX44) based on Intel Xeon MP processors with a clock frequency of up to 2.80 (3) GHz. The thickness of such a module was two times greater than that of the dual-processor configuration announced in the fall of 2003. Up to seven of these modules are installed in the GEG Express Blade universal server chassis. While maintaining the density of the solution - up to 28 processors in a single 19-inch chassis with a height of 7U - the use of four-processor modules allows you to get a significant increase in overall system performance due to the transition to processors with MP architecture.

The new module is equipped with four Gigabit Ethernet controllers and allows the installation of two ATA 100 or Ultra320 SCSI hard drives (using a special external drive cage mounted on the module), which can be combined in RAID level 0 or 1. In addition, the new module can be equipped with additional "fast" "Interfaces such as Fiber Channel and Myrinet.

With the advent of the new four-processor implementation of the modules, the GEG Express Blade system, thanks to its highly integrated design, demonstrates excellent volume / performance and price / performance ratios.

With high specific productivity per unit volume, the compact GEG Express Blade systems allow customers with large server parks to more efficiently use the available space and significantly save both work space and energy consumption. Managing all GEG Express Blade server modules from a single console significantly reduces system administration costs. Due to the high performance of the GEG Express Blade system based on four-processor modules when working with large information arrays (for example, with large databases), it can be successfully used in research institutions, companies in the telecommunications, banking and manufacturing industries.

A distinctive feature of Kraftway GEG Express Blade is the ability to combine two and four-processor modules in one chassis and install different OS on different modules depending on the tasks performed. All released modules are certified for compatibility with Microsoft Windows 2003 Server and Linux Red Hat. GEG Express Blade servers come with a five-year warranty, and during the first two years the system is serviced on-site.

Intel Deployment Manager software allows you to automate many administrative tasks related to monitoring systems, installing software, and various updates to BIOS, Firmware, etc. blade servers using an intelligent software environment. You can, for example, automate the following tasks:

• detection of changes in hardware and software configuration;
• tracking changes in software licenses (allows you to effectively manage licenses - add them, delete, move);
• simultaneous software installation on multiple servers;
• bulk update of BIOS and Firmware versions;
• effective management of local network settings, logical division into groups according to predefined signs, differentiation of user rights to manage the entire complex;
• manage all blade server resources from one place.

Traditionally, each of the above tasks is performed by different groups within the organization. As soon as the group has completed one task, another is taken for work, and so on, until all the necessary settings and installations are completed. Intel Deployment Manager allows you to fully automate such processes, eliminating all manual operations.

Today, the company’s product line also includes GEG Express Blade-N, a 3U module-class system containing up to 18 server modules with hot-swappable and common control. It is primarily intended for use in high-density computing systems. When mounting a GEG Express Blade-N in a standard 42U server cabinet, the system will house 252 server modules with Intel Pentium III processors, 100/1000 Mbps network interfaces, hard drives and standard PCI expansion slots.