Last year I wrote a short blog post looking at HP ProLiant servers and how the model line-up looks.  I haven’t looked at IBM System x for a few years but last week I got the chance to sit down and have a look at the current Dell PowerEdge portfolio.

Just as for HP, there is some logic behind Dell’s server names, although this scheme is fairly new and some older servers (e.g. the PowerEdge 2950) do not fit this:

• The first character is a letter indicating the chassis type: T for tower; R for rack; M for modular (blade).
• The next digit indicates the market segment for which the server is destined: 1, 2 and 3 are single-socket servers for value, medium and high-end markets respectively; 4 and 5 are 2 socket value servers with 6 for medium and 7 for high-end; 8 indicates a special server (for example one which can be configured as a 2 or a 4-socket machine); 9 indicates a 4 socket server (Dell does not currently complete in the 8-way marketplace).
• The next digit indicates the generation number (0 for 10th, 1 for 11th, 2 for 12th generation). With a new generation every couple of years or so, resetting the clock to zero should give Dell around 20 years before they need to revisit this decision!
• Finally, Intel servers end with 0 whilst AMD servers end with 5.

There is another complication though – those massive cloud datacentres operated by Microsoft, Amazon, et al use custom servers – and some of them come from Dell.  In that scenario, the custom servers don’t need to be resilient (the cloud provides the resilience) but Dell has now brought similar servers to market for those who want specialist, high-volume servers, albeit with a slightly lower MTBF than standard PowerEdge boxes.  So, for example: the C1100 is a 2-way, 1U server that can take up to 18 DIMMs for memory-intensive applications; the C2100 is a 2-way, 2U server with room for 12 disks (and 18DIMMs); whilst the C6100 crams four 2-way blades into 2U enclosure, with room for 12 DIMMs and up to 24 2.5″ disks!