back to article IBM racks up the memory

While wandering the floor at the recently concluded VMworld 2010, we stopped by the IBM booth to take a look at their wares. We ran into an old pal of ours, Bob Zuber, and he walked us through their MAX 5 memory extension feature. IBM racks up the memory var MediaPlayerVars = { '_': { 'http://www.youtube.com/v/9hjVtRQyYkY …

COMMENTS

This topic is closed for new posts.
  1. Anonymous Coward
    Megaphone

    Dimm View

    I can see the point in having lots of cheap memory as opposed to lots of big expensive DIMMs in the server, but I have a gut feeling that the cost of the memory board coupled with the price that IBM will charge for its own DIMMs will make the whole product ridiculously expensive and only viable for those that need vast amounts of memory at any cost.

    If on the other hand I can use cheap third party DIMMs this becomes a very attractive proposition. But again my gut feeling tells me this will "void my warranty"

    1. danolds
      Thumb Up

      Good point..

      ...it really depends on the terms and conditions attached to the deal, right? Along with the price of the memory drawer too. Right now, it's hard to pry those details out of them without buying one myself. While I'm committed to providing as much info as I can, I'm not going to take one for the team to that extent...lol

      I do seem to remember IBM and other vendors talking about how they've dropped their memory prices in past years. I'm sure that those prices aren't competitive with the cheapest memory out there - but are the street prices from the vendors competitive with what you'd get from a name brand memory house?

    2. Matt Bryant Silver badge

      RE: Dimm View

      ".....But again my gut feeling tells me this will "void my warranty"" Yes, but then that's the same for all our vendors. We make a point of putting only the vendors' parts into our servers because we did once get caught out by a reseller here in the UK, who was putting third-party memory into the boxes and we only found out when they used a bad batch. However, memory is a bit like that old joke about scoping costs - "guesstimate as closely as you can as to what you'll need, then double it and add a bit more and you may be just about close"! We will accurately scope the initial server requirement, we'll get it in and it will all run fine, and then someone will add a patch or a new version of the application with a new (memory-hungry) feature, and suddenly we're looking for extra slots to stick memory into. I always try and get our servers with the biggest DIMMs going because I know that means I can have slots free for the inevitable unexpected memory upgrades.

  2. Morten Bjoernsvik
    Thumb Down

    performance?

    I've seen older IBM X-boards and they all have terrible bandwidth (PCIX?). Not much better than running SD cards in parallell.

    1. Tom Samplonius
      Thumb Up

      Re: performance?

      @Morten: "I've seen older IBM X-boards and they all have terrible bandwidth (PCIX?). Not much better than running SD cards in parallell."

      I don't think you can get PCI-X equipment servers from anyone anymore, let alone IBM. But PCI-X 64bit 133Mhz had bandwidth of about 1000MB/s, so it still pretty fast for storage, and certainly far faster than "SD cards in parallel". Plus, many servers of that generation had multiple PCI-X buses too.

      However, other than servers, most systems only came with 32bit PCI-X slots. Or set the 64bit slots to 32bit mode by default, unless re-enabled in the BIOS. So user issues contributed to the speed issues. But with 64bit cards in 64bit slots, PCI-X was fast for its time. And still pretty fast today.

    2. danolds

      Not PCIX

      They are connecting the memory drawer to the host system via Intel QPI and what IBM used to call their Scalability Port (they might still call it that). This port is the same port they used to use to pin 4-socket chassis together to build out 16-socket SMP systems. I don't think they ever published full specs on this port, but it was fast and wide enough to give them reasonable SMP scalability on commercial apps.

This topic is closed for new posts.

Other stories you might like