Choosing a problem: Analyzing the difference between a blade server and a rack server

In recent years, the server market is the most popular http://www.aliyun.com/zixun/aggregation/9785.html "> server architecture than the blade server, it appears in a large number of media, Major manufacturers have also launched a variety of corresponding models. What advantages does a blade server have over a traditional server? What should consumers choose when choosing a server? This paper compares the blade server with the traditional rack-type server, and introduces the experience of the two in the selection.

In the comparison of Blade server and rack server, we mainly focus on the investment cost, performance configuration, expansion capability, energy consumption, maintenance and management of the products. The following is a concrete example to compare. This example is IBM's HS21 and HS21 XM two blade servers, as well as configured x3550, x3650 rack-like servers.

The HS21 and HS21 XM Two blades are chassis with IBM Blade Center H (bc-h). The Bc-h chassis consists of a 10GB Ethernet switch, a 1GB Ethernet switch, and four power supply systems. The XM Series Blades are equipped with a pair of 2.33GHz Intel four core processors, while the HS21 CPU frequency of 2.0GHZ;XM is 16GB DDR2, while HS21 is only 8GB.

The above configuration and IBM x3550 rack-type server is not very different, but the frequency increased, for 2.66GHz. IBM x3650 is a 2U dual quad-core rack server with a maximum processor frequency of 3.0GHz.

If the knife version of the chassis installed 14 "Blades", the total cost of installing the HS21 "Blade" is:, 100*14+,000=6,400; if the HS21 XM "Blade" is installed, the total cost is:, 800*14+,000=4,200.

And the total cost of 14 sets of x3550 is:, the total cost of 900*14=0,600;14 x3650 is:, 300*14=0,200. Note that the final total cost also calculates the cost of a 24-port 1G Fibre Ethernet switch, about 500. Then the total cost of the above two kinds of machine-flail server scheme is changed into: 3,100 and 2,700 respectively. Compare the above two kinds of blade server total cost rise to know, 14 HS21 "Blade" blade server Total cost than two 14 rack-type server solution is cheaper. The total cost of a blade server with 14 HS21 XM "Blades" is comparable to that of the expensive 14-x3650-Rack server scheme.

Testing by a number of professional testing agencies shows that one of the most demanding tests for the server is--processor FORK+EXECVE (primarily test data transfer speed between I/O and LMBench3), H21 and h.21 The XM Blade Server test results are 289.9 microseconds, while the x.3550 rack server is 173.2 microseconds, up 40% faster (longer, lower speed). This is because the h.21 and h.21 XM Blade servers are configured to be RAID1, and X.3650 is configured on the rack-x.3550 and RAID5 servers. Other aspects of the test structure show that the performance of these servers are comparable, because the unit configuration of these servers are basically the same, the use of two four core Intel processors, in the frequency test also chose the same configuration. As a result, there is not much difference in performance between blades and rack servers, and the main difference is only in the data access capabilities that are affected by the disk array.

Scaling capability Comparison

In terms of capacity expansion, we have to compare two aspects. One is a comparison of a single server, and the other is a comparison of server cluster extensions.

The single server in the blade server is each "blade". In these "blades" because of their high density determines their ability to expand very limited. This is not only the advantages of blade server, but also the inevitable shortcomings. The generic "blade" is equipped with only two hard drives, with a maximum of RAID 1 mirroring, or RAID 0 without redundant performance hard disk arrays, which means that the hard disk capacity cannot exceed the maximum capacity specified by the blade, nor can it maximize server disk read/write performance. This is the blade server we described earlier processor FORK+EXECVE test results are far less than the reason for the rack-like server.

Unless the server installs and configures iSCSI, Fibre Channel, or other SAN connections, it is difficult for blades to substantially improve disk capacity and read and write performance.

Unlike blades in a blade server, the rack-type server has a much larger expansion space, and the sample can also be configured with more components. A 2U x3650 rack server has a built-in RAID 5 with four hard drives: three make up a raid and a hot backup hard drive. Some vendors even placed 1U of rack-type servers with equivalent 2U of storage capacity devices.

In addition to the disk expansion capabilities, in memory, I/O expansion slots, the same phenomenon, that is, Blade server "blade" scalability than the rack-type server.

In addition, there is another potential downside to the blade server, which is its proprietary nature. If you choose a manufacturer's blade server, it means that the service item, the component Update, is to be obeyed by the vendor, and the services and components of other vendors are not available. And, in general, future updates, faster CPUs will be first used on the rack server.

With the ability to scale clusters, blades have the innate advantage of simply inserting each "blade" into the blade chassis, and using their own cluster management software to easily configure the blade server farm. A rack-like server is not as easy to build as a cluster, it must buy additional cluster management software, and network connectivity, configuration is more complex.