Hey folks,
We've got some new hardware and are trying to figure out what best to do with it. Either run CentOS right on the bare metal, or virtualize, or several combination options. Mainly looking at :
- CentOS on bare metal - CentOS on ESXi 4.0 with local disk - CentOS on ESXi with 1 VM running Openfiler to serve disk to other VMs
And want to benchmark these 3 scenarios
So far all we have is a dd-based disk IO benchmark.
What else can you all recommend.
BTW, we also ideally want to try each of the above with a Postgres DB as well (and once without)
thanks, -Alan
On 03/11/2009, at 11:52 PM, Alan McKay wrote:
Hey folks,
We've got some new hardware and are trying to figure out what best to do with it. Either run CentOS right on the bare metal, or virtualize, or several combination options. Mainly looking at :
- CentOS on bare metal
- CentOS on ESXi 4.0 with local disk
- CentOS on ESXi with 1 VM running Openfiler to serve disk to other
VMs
And want to benchmark these 3 scenarios
So far all we have is a dd-based disk IO benchmark.
What else can you all recommend.
If IO testing is your primary concern then bonnie++ would be very useful to get good reports on your different configurations. It can generate nice HTML formatted reports for you as well.
Regards, Oliver
BTW, we also ideally want to try each of the above with a Postgres DB as well (and once without)
thanks, -Alan
-- “Don't eat anything you've ever seen advertised on TV” - Michael Pollan, author of "In Defense of Food" _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
I'd recommend a quick read through this post:
http://lethargy.org/~jesus/writes/disk-benchmarking-with-dd-dont
And then browse through the docs for these tools and pick the one that suites your needs best:
http://www.iozone.org/ http://sysbench.sourceforge.net/ http://www.coker.com.au/bonnie++/
-ryan
On Tue, Nov 3, 2009 at 8:22 AM, Alan McKay alan.mckay@gmail.com wrote:
Hey folks,
We've got some new hardware and are trying to figure out what best to do with it. Either run CentOS right on the bare metal, or virtualize, or several combination options. Mainly looking at :
- CentOS on bare metal
- CentOS on ESXi 4.0 with local disk
- CentOS on ESXi with 1 VM running Openfiler to serve disk to other VMs
And want to benchmark these 3 scenarios
So far all we have is a dd-based disk IO benchmark.
What else can you all recommend.
BTW, we also ideally want to try each of the above with a Postgres DB as well (and once without)
thanks, -Alan
-- “Don't eat anything you've ever seen advertised on TV” - Michael Pollan, author of "In Defense of Food" _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
http://lethargy.org/~jesus/writes/disk-benchmarking-with-dd-dont
BTW, while I'll admit up front that Theo is probably a lot better at this stuff then I will ever hope to be (just starting to read his book), I don't see a reason in that post as to why not. Other than "because I said so". He runs some tests and assumes we know how to draw the same conclusions from them that he does, but does not really explain what those conclusions are. Now go back to my first clause of my first sentence .
Not that I would think of using this as my only benchmark - I will still be doing the dd benchmarks.
On Tue, Nov 03, 2009 at 09:22:57AM -0400, Alan McKay wrote:
Hey folks,
We've got some new hardware and are trying to figure out what best to do with it. Either run CentOS right on the bare metal, or virtualize, or several combination options. Mainly looking at :
- CentOS on bare metal
- CentOS on ESXi 4.0 with local disk
- CentOS on ESXi with 1 VM running Openfiler to serve disk to other VMs
And want to benchmark these 3 scenarios
So far all we have is a dd-based disk IO benchmark.
What else can you all recommend.
BTW, we also ideally want to try each of the above with a Postgres DB as well (and once without)
If you want to test IO performance I recommend vdbench: http://przemol.blogspot.com/2008/06/vdbench-disk-io-workload-generator.html
Regards Przemyslaw Bak (przemol) -- http://przemol.blogspot.com/
------------------------------------------------------------------------------ Strrraszny konkurs z Scooby-Doo! Wez udzial >> http://link.interia.pl/f2412
Alan McKay wrote:
Hey folks,
We've got some new hardware and are trying to figure out what best to do with it. Either run CentOS right on the bare metal, or virtualize, or several combination options. Mainly looking at :
- CentOS on bare metal
- CentOS on ESXi 4.0 with local disk
- CentOS on ESXi with 1 VM running Openfiler to serve disk to other VMs
And want to benchmark these 3 scenarios
So far all we have is a dd-based disk IO benchmark.
What else can you all recommend.
BTW, we also ideally want to try each of the above with a Postgres DB as well (and once without)
You can try fsbench which is not a generic benchmark. However, you mention Postgres DB so it may or may not be useful to you. fsbench simulates delivery to a maildir and simulates a single writer/reader to 16 writers/readers. Of course, it uses fsync calls unlike certain benchmarking software that do nothing of the sort like postmark. If you need a copy of fsbench, I have the original tarball that Bruce Guenter published.
Alan McKay wrote:
Hey folks,
We've got some new hardware and are trying to figure out what best to do with it. Either run CentOS right on the bare metal, or virtualize, or several combination options. Mainly looking at :
- CentOS on bare metal
- CentOS on ESXi 4.0 with local disk
- CentOS on ESXi with 1 VM running Openfiler to serve disk to other VMs
And want to benchmark these 3 scenarios
Save yourself some trouble and skip the last one. Running openfiler in a VM to serve disk to other VM's is a crazy idea.
Save yourself more trouble and think about what your going to use the system for, if it is(or will be) a multi purpose system, then go with ESXi, if it's a single purpose system go with bare metal.
ESX has maximum efficiency with 1 vCPU VMs, so if you have 8 cores, you should have at least 8 VMs running.
http://portal.aphroland.org/~aphro/vmware/09Q3-perf_overview_and_tier1-pac_n...
nate
On Nov 3, 2009, at 8:22 AM, Alan McKay alan.mckay@gmail.com wrote:
Hey folks,
We've got some new hardware and are trying to figure out what best to do with it. Either run CentOS right on the bare metal, or virtualize, or several combination options. Mainly looking at :
- CentOS on bare metal
- CentOS on ESXi 4.0 with local disk
- CentOS on ESXi with 1 VM running Openfiler to serve disk to other
VMs
And want to benchmark these 3 scenarios
So far all we have is a dd-based disk IO benchmark.
What else can you all recommend.
BTW, we also ideally want to try each of the above with a Postgres DB as well (and once without)
Take a look at iozone it is a little dated, but still good and is cross platform so you can can have a level playing field between Linux, Windows and Solaris.
I'd be interested in hearing your results. I will tell you it will be slower then bare metal, that is just a fact of life, but with adequate hardware it may be more then acceptable.
It would allow a VM to fail-over for another though which is key in a HA situation like storage.
-Ross
Ross Walker wrote:
On Nov 3, 2009, at 8:22 AM, Alan McKay alan.mckay@gmail.com wrote:
Hey folks,
We've got some new hardware and are trying to figure out what best to do with it. Either run CentOS right on the bare metal, or virtualize, or several combination options. Mainly looking at :
- CentOS on bare metal
- CentOS on ESXi 4.0 with local disk
- CentOS on ESXi with 1 VM running Openfiler to serve disk to other
VMs
And want to benchmark these 3 scenarios
So far all we have is a dd-based disk IO benchmark.
What else can you all recommend.
BTW, we also ideally want to try each of the above with a Postgres DB as well (and once without)
Take a look at iozone it is a little dated, but still good and is cross platform so you can can have a level playing field between Linux, Windows and Solaris.
I'd be interested in hearing your results. I will tell you it will be slower then bare metal, that is just a fact of life, but with adequate hardware it may be more then acceptable.
Odds are pretty good that the limiting factor will be disk head seek time regardless of the software layers above, although this may not show up in single-tasking benchmarks like it does in typical use. The disk/partition layout and the way competing tasks are distributed across them are likely to have more effect on speed than an extra logical disk software wrapper.
On Nov 3, 2009, at 11:16 AM, Les Mikesell lesmikesell@gmail.com wrote:
Ross Walker wrote:
On Nov 3, 2009, at 8:22 AM, Alan McKay alan.mckay@gmail.com wrote:
Hey folks,
We've got some new hardware and are trying to figure out what best to do with it. Either run CentOS right on the bare metal, or virtualize, or several combination options. Mainly looking at :
- CentOS on bare metal
- CentOS on ESXi 4.0 with local disk
- CentOS on ESXi with 1 VM running Openfiler to serve disk to other
VMs
And want to benchmark these 3 scenarios
So far all we have is a dd-based disk IO benchmark.
What else can you all recommend.
BTW, we also ideally want to try each of the above with a Postgres DB as well (and once without)
Take a look at iozone it is a little dated, but still good and is cross platform so you can can have a level playing field between Linux, Windows and Solaris.
I'd be interested in hearing your results. I will tell you it will be slower then bare metal, that is just a fact of life, but with adequate hardware it may be more then acceptable.
Odds are pretty good that the limiting factor will be disk head seek time regardless of the software layers above, although this may not show up in single-tasking benchmarks like it does in typical use. The disk/partition layout and the way competing tasks are distributed across them are likely to have more effect on speed than an extra logical disk software wrapper.
Disk head seek time is always the limiting factor, but partition misalignment can cause significant artificial limits if blocks overlap page size boundaries and the next block and maybe chunk needs to be read/written.
I have seen it limit random throughput by as much a 5 times on small blocks.
-Ross
On Tue, Nov 3, 2009 at 6:43 PM, Ross Walker rswwalker@gmail.com wrote:
On Nov 3, 2009, at 11:16 AM, Les Mikesell lesmikesell@gmail.com wrote:
Ross Walker wrote:
On Nov 3, 2009, at 8:22 AM, Alan McKay alan.mckay@gmail.com wrote:
Hey folks,
We've got some new hardware and are trying to figure out what best to do with it. Either run CentOS right on the bare metal, or virtualize, or several combination options. Mainly looking at :
- CentOS on bare metal
- CentOS on ESXi 4.0 with local disk
- CentOS on ESXi with 1 VM running Openfiler to serve disk to other
VMs
And want to benchmark these 3 scenarios
So far all we have is a dd-based disk IO benchmark.
What else can you all recommend.
BTW, we also ideally want to try each of the above with a Postgres DB as well (and once without)
You can try out stress. There is a package for it from rpmforge.
Matt
-- Mathew S. McCarrell Clarkson University '10
mccarrms@gmail.com mccarrms@clarkson.edu 1-518-314-9214
Just gonna bump this because I don't see any CPU benchmark recommendations yet.
Anyone?
Thanks for all the responses so far!
On Tue, Nov 3, 2009 at 8:22 AM, Alan McKay alan.mckay@gmail.com wrote:
Hey folks,
We've got some new hardware and are trying to figure out what best to do with it. Either run CentOS right on the bare metal, or virtualize, or several combination options. Mainly looking at :
- CentOS on bare metal
- CentOS on ESXi 4.0 with local disk
- CentOS on ESXi with 1 VM running Openfiler to serve disk to other VMs
And want to benchmark these 3 scenarios
So far all we have is a dd-based disk IO benchmark.
What else can you all recommend.
BTW, we also ideally want to try each of the above with a Postgres DB as well (and once without)
thanks, -Alan
-- “Don't eat anything you've ever seen advertised on TV” - Michael Pollan, author of "In Defense of Food"
Alan McKay wrote:
Just gonna bump this because I don't see any CPU benchmark recommendations yet.
Plenty of results for you to reference against from a massive array of vendors.
nate
-
Alan McKay -
Chan Chung Hang Christopher -
Les Mikesell -
Mathew S. McCarrell -
nate -
Oliver Ransom -
przemolicc＠poczta.fm -
Ross Walker -
Ryan Cox