1

We have an IBM x3650 M2 server. It has 2 Intel X5570 CPUs (quad-core Nehalem). The current memory configuration is 12x 4GB DIMMs for a total of 48GB, running at 1066MHz.

To increase the amount of memory to 72GB I can populate the 4 free slots with 4GB DIMMs, but this has the side effect of dropping the memory speed down to 800MHz. Our vendor is strongly advising against this, and suggests replacing some of the 4GB DIMMs with 8GB DIMMs in order to maintain the memory speed at 1066MHz.

I've got 4x 4GB DIMMs on loan from the vendor, so I'm looking for a way to quantify the difference.

The box is running ESXi 4.1 and it's in a cluster with 2 other ESXi hosts, both of which are newer X3650 M3 boxes.

What can I do to test whether the change in memory speed has a significant impact? If it matters, I can take this host out of the cluster for the duration of testing.

EDIT: I should've been clearer.... I would like to determine if there's any impact on running VMs, so installing a "real" OS or using memtest86 won't really do much to further my goal.

4 Answers 4

1

The impact is greater than if you were running an older CPU with no support for EPT, but realistically the only way to determine if it's going to affect your workload is to actually profile the workload. Don't take the hypervisor out of the equation, because the whole point is to test the setup that you're running, not some hypothetical benchmark figure. Ignore Memtest86+, ignore bare-metal OSes, just find a virtual machine that's representative of a memory-intensive workload in your environment and beat the crap out of it.

Guy is dead-on: most consolidated workloads are bound by the amount of memory in the system and not by any other resource. By decreasing memory contention, the extra memory will probably help you out by enabling you to use more of your memory-intensive VMs' RAM as cache instead of it being ballooned out under pressure.

1

When you use vmware you sacrifice "outright speed" for consolidations of servers.

Unless you have processes that are totally RAM MHz speed dependant then I would suggest that the extra RAM will IMPROVE performance.

Why? If vmware is short of RAM then it will start to use memory ballooning / stealing techniques to redistribute RAM to where it is mostly needed. This has an impact on performance for those servers on the sharp end of the redistribution.

If you add more total RAM, then the memory rationing will go away and overall performance will improve.

Therefore:

1/ Are you REALLY in need of more RAM. If not, then adding the extra sticks you will get a 20% drop is RAM performance for every server on the host.

2/ If your server is already overcommitted and is in "memory management mode" (typically if RAM is over 80% utilised) then the extra RAM will improve OVERALL performance for all the servers.

Obvioulsy, the proper answer is to swap out ALL the RAM for larger same type DIMM's.

1
  • That's just the thing, I don't honestly think that a drop from 1066MHz to 800MHz will have a significant impact. It's nice to have something to back that up though, as gut feelings don't generally sway decisionmakers either way. :-) Nov 9, 2010 at 13:08
0

Take server out. Install "real" os.

Check http://www.tomshardware.com/reviews/ram-speed-tests,1807.html for approeaches to measure the impact of memory speed. That pretty much is the only approach you have.

0

You could boot the server up using MEMTEST+, that not only tests your memory but gives you some hopefully good-enough overall speed figures, it's quick and low impact.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .