IBM System x

The IBM System x computers form a sub-brand of International Business Machines (IBM's) System brand servers (the other System sub-brands having the names IBM System i, IBM System p, IBM System z and IBM System Storage). In addition IBM System x is the main component of the IBM System Cluster 1350 solution.

History

Starting out as IBM PC Server, rebranded Netfinity, then eServer xSeries and now System x, these servers are distinguished by being based on off-the-shelf x86 CPUs; IBM positions them as their "low end" or "entry" offering.

Previously IBM servers based on AMD Opteron CPUs did not share the xSeries brand; instead they fell directly under the eServer umbrella. However, current AMD Opteron-based servers fall under the System x brand.

IBM PC Server

PC Server range

• IBM PC Server 300
• IBM PC Server 310 (PCI/ISA)
• IBM PC Server 315 (PCI/ISA)
• IBM PC Server 320 (PCI/EISA)
• IBM PC Server 325 (PCI/EISA)
• IBM PC Server 330 (PCI/EISA)
• IBM PC Server 500 (MCA)
• IBM PC Server 520 (PCI/EISA or PCI/MCA)
• IBM PC Server 704 (PCI/EISA)
• IBM PC Server 720 (PCI/MCA)

Numbering scheme

• 300 range for high-volume, entry level servers
• 500 range for midrange
• 700 range for high-end.

IBM Netfinity

Not to be confused with a different IBM product with a similar name, NetFinity (notice the capital F).

Netfinity range

• IBM Netfinity 1000
• IBM Netfinity 3000, 3500
• IBM Netfinity 4000R, 4500R
• IBM Netfinity 5000, 5100, 5500, 5600
• IBM Netfinity 6000R
• IBM Netfinity 7000, 7000-M10, 7100, 7600
• IBM Netfinity 8500R

Numbering scheme

The numbering scheme started off similar to that of the IBM PC Servers, but additional ranges were added, like the entry-level 1000 later on. Models ending with an R, are rack-mount.

KVM Cabling scheme

Some Netfinity servers used IBM's C2T cabling scheme for Keyboard/Video/Mouse.

IBM eServer xSeries

xSeries range

• IBM eServer xSeries 100, 130, 135, 150
• IBM eServer xSeries 200, 205, 206, 206m, 220, 225, 226, 230, 232, 235, 236, 240, 250, 255, 260
• IBM eServer xSeries 300, 305, 306, 306m, 330, 335, 336, 340, 342, 345, 346, 350, 360, 365, 366, 370, 380, 382
• IBM eServer xSeries 440, 445, 450, 455, 460

Numbering scheme

• 100 series are entry-level tower servers
• 200 series are tower servers
• 300 series are rack-mount servers
• 400 series are rack-mount scalable servers

KVM Cabling scheme

Many xSeries servers used IBM's C2T cabling scheme for Keyboard/Video/Mouse.

IBM eServer

eServer range

• IBM eServer 325, 326, 326m
• IBM eServer BladeCenter, BladeCenter T, BladeCenter H, BladeCenter HT

Numbering scheme

For marketing reasons the AMD processor based e325, e326 and e326m and the BladeCenter which supports non-Intel processor products were not branded xSeries, but were instead placed directly under the eServer brand. The xSeries brand was limited to only Intel-based server products.

From a numbering perspective the AMD servers did fit into the xSeries range, under the similar x335 and x336 Intel processor products. These numbers were not re-used in the xSeries range to prevent confusion.

IBM System x

IBM System x range

• IBM System x3105, 3100
• IBM System x3200, x3200 M2, x3250, x3250 M2, x3250 M3
• IBM System x3350
• IBM System x3400, x3450, x3455
• IBM System x3500
• IBM System x3550, x3550 M2, x3550 M3
• IBM System x3650, x3650T, x3655, x3650 M2, x3650 M3
• IBM System x3755
• IBM System x3800, x3850, x3850M2
• IBM System x3950, x3950 M2

Enterprise eX5 architecture

IBM eX5 systems innovations introduced on March 2, 2010 include:

• MAX5 – Unprecedented memory expansion with the external MAX5 memory chassis, decoupling server memory from system processors to allow you to optimize server performance
• eXFlash – Get dramatically faster I/O, with greater storage density and improved reliability
• FlexNode – Gain the ability to re-deploy server on a project-by-project basis for superior asset utilization and workload management
• Scalability – Updating IBM’s innovative ‘pay-as-you-grow’ capability by enabling all eX5 servers to scale
• Workload optimized systems – Delivering specific workload-tuned models to minimize your time to value by leveraging IBM’s industry expertise

Enterprise X4 architecture

In large, scalable servers, the importance of the memory, I/O, and disk subsystems are paramount. To efficiently utilize multiple execution engines, whether they are cores, threads, or physical processors, two things need to be achieved.

First, the application must have the ability to distribute workload across the execution engines, this applies to the hypervisor, the operating system, and the application. Second, the memory and I/O subsystems must have the physical ability to feed the execution engines with enough data to keep them fully utilized. These two requirements are easier said than done; indeed the average processor utilization in an x86 computing platform is 5 - 10%.

Software vendors have previously been able to fall back on the increasing clock speed of the processors for performance gains, however the recent shift to multi-core processing requires intelligence coded into the applications to support parallel processing. The applications can eventually be recoded, but this takes time and a finite limit always exists to the performance benefits that can be achieved by adding processor cores. Using software to virtualize your servers offers a way to use existing applications, not written for parallel processing, and still scale up in performance because the hypervisor can distribute multiple virtual machines across all of the processor cores in the system.

Virtualization brings many other benefits such as increased flexibility for server fault tolerance, disaster recovery, and power savings that reduce costs and support environmental responsibility. It is however a software solution and can therefore do little to alleviate the problem of physically feeding data into the processing units fast enough to maximize utilization. The performance of the processor, memory and I/O subsystems inside an x86 server must complement each other to allow virtualization software to get the most use out of the machine. The System x data center is built to face current industry challenges.

Numbering scheme

2nd digit increments to show capability

3rd digit is a 0 for tower models, and 5 for rack-mount

4th digit is a 0 for Intel processors, and 5 for AMD Opteron.

Models with a T at the end are meant for Telco purposes.

iDataPlex

An iDataPlex rack in a Portable Modular Data Center.

• SciNet Consortium's IBM System x iDataPlex Supercomputer^[1][2][3]

See also

• List of IBM products

References

1. ^ "Toronto team completes Canada's most powerful supercomputer". CBC News. June 18, 2009. http://www.cbc.ca/technology/story/2009/06/18/tech-090618-ibm-supercomputer-scinet-toronto.html. Retrieved 2009-06-18. 
2. ^ Hall, Joseph (June 18, 2009). "U of T supercomputer probes origins of the universe". The Star. http://www.thestar.com/news/gta/article/652745. Retrieved 2009-06-18. 
3. ^ "University of Toronto's Supercomputer Goes Online Thursday". All Headline News. June 18, 2009. http://www.allheadlinenews.com/articles/7015533080. Retrieved 2009-06-18. 

External links

• IBM's System x
• IBM's server homepage
• IBM System x Twitter, Blogs and Commumities
• IBM System x Redbooks
• IBM System x Configuration and Options Guide for current systems
• Tuning IBM System x Servers for Performance - IBM Redbooks
• IBM Netfinity Reference
• IBM PC Servers Reference
• Real World Technologies: x86 Servers Brace for a Hurricane an in-depth overview of the X3 architecture, covering the cache coherency protocol, snoop filter, memory and I/O subsystems and the remote directory, by David Kanter
• Community Portal with the latest news on IBM System x, BladeCenter and Systems Management
• The Price of Safety: Evaluating IOMMU Performance (mentions the Calgary IOMMU which is featured in eSeries AMD Opteron systems.
• IBM's System x configurator and system builder