=== zz_bigbash is now known as bigbash
bigbashping h00k00:10
=== bigbash is now known as zz_bigbash
=== zz_bigbash is now known as bigbash
=== bigbash is now known as zz_bigbash
=== zz_bigbash is now known as bigbash
bigbashh00k, what processor do you use in your vSphere setup?14:09
CheeseheadThe EMEA Membership board is meeting at 2000 UTC today (five hours from now) in #ubuntu-meeting. Anyone who wants to support (or oppsose) my Ubuntu Membership application is welcome!15:09
bigbashCheesehead, at 2CT or 3CT?15:53
Cheeseheadbigbash: 2PM CST16:53
bigbashok sounds good16:54
h00kbigbash: oh hi20:23
h00kshooot, I was late.20:24
h00kI don't think there'll be any issues20:24
bigbashI don't think so20:25
bigbashlots to vouch for him20:25
bigbashCongrats Cheesehead!!20:25
h00kCheesehead: congrats!20:25
CheeseheadThanks everybody!20:25
CheeseheadWait...I said that in the other channel.20:25
CheeseheadThanks FOR TAKING THE TIME TO COME BY!20:26
CheeseheadThanks for your testimonials and efforts.20:26
h00kI'm working from home today20:29
h00kbigbash: coincidentally, vmware vcp test tomorrow.20:29
bigbashh00k, nice20:29
h00kbigbash: we have a pretty nice environment set up, although I wouldn't recommend the Cisco UCS 250's yet20:39
h00kthere's some firmware issues that need to get fixed. It's first-round hardware :(20:40
bigbashsorry helping someone in another channel20:44
h00kno prob20:45
bigbashyou think that 1 host could run 10-20 vm?20:45
h00kwhat are the VMs doing?20:45
h00kThat's the biggest part, really.20:45
bigbashI'm just being super cautious because I have 1 vendor we've worked with forever saying Xeon and the other with the HP engineers saying AMD20:45
h00kI am personally an Intel fan, but I really don't have any solid evidence towards one or the other20:47
bigbashWebserver, DC/File/Print, Payroll, Accounting, Exchange, and old SNA server to corporate, and a reporting server that has 10 ppl RDP20:47
h00kMemory is the biggest thing, really.20:47
bigbashthat's what it really sounds like20:47
h00kbigbash: cool, with that license, you can do HA for things. That's nice20:47
bigbashthe combined amount of RAM all 7 have is like 40GB20:48
bigbashyea the HA is what I was after20:48
h00kbigbash: cool.  One thing you can do too, to estimate how much 'stuff' you'll need, is turn on PerfMon for a week on your servers and collect data on IOPS20:49
bigbashhmmm thats a good idea20:49
bigbashEgyParadox, you there?20:50
bigbasharen't you supposed to be in #ubuntu-meeting?20:50
h00kbigbash: and they have guided consolition20:52
bigbashnice, I've been plowing through videos and readings getting ready20:53
h00kbigbash: http://goo.gl/2sgBg20:54
bigbashI've been waiting to change over to virtual20:54
bigbashdown the road we are going to test thin clients but we'll add more hosts at that point20:55
h00kbigbash: what are you using for storage?20:57
h00kbigbash: also: http://whughgriffin.wordpress.com/2009/10/16/vmware-vsphere-4-guided-consolidation/20:57
bigbashwe are using the HP P4300 Starter SAN http://h10010.www1.hp.com/wwpc/us/en/sm/WF06b/12169-304616-3930449-3930449-3930449-4118659-4118705-4118707.html21:05
bigbashit's pretty much a full replacment of infrastructure21:05
h00kbigbash: cool, we just spec'd out a P2000 G3 for a client21:07
bigbashdoubt we need the whole 7.2TB21:08
bigbashI think we are using about 1.5-2Tb currently21:08
bigbashI saw that VMWare Standard Acceleration comes with SLES21:09
h00kbigbash: oh, interesting.21:10
h00kbigbash: are you doing ESX or ESXi21:10
h00kbigbash: also beware that the more virtual processors that you give a virtual machine, it won't necessarily speed it up21:11
bigbashif it doesn't speed it up what does it do?21:12
bigbashESXi is the only hypervisor available in vSphere 521:12
h00kphew, good.21:13
bigbashso I guess that is what we are going with21:13
h00kI thought so.21:13
h00kSo, if you give it >1 processor, it can actually slow it down if it wants access to both cores.  If you have machines wanting more cores, you can create contention because they're going to want them open at the same time21:13
h00kIf that makes sense21:14
bigbashnot exactly but I can read up that part more, so more cores is better for processor then?21:15
h00keach core, and if you have hyperthreading, a core will show up as 2 processors21:15
h00kfor instance, our dual hex cores with hyperthreading show as 2x6x2 = 24 processors available to VMs21:15
h00kA two-vCPU virtual machine has to wait for two physical processors to free up, but the hypervisor doesn’t like to have idle CPUs, so it runs a single vCPU virtual machine instead. It ends up being a long time before two physical CPUs free up simultaneously, and the two vCPU virtual machine seems really slow as a result. By “seems really slow” I mean it doesn’t perform very well, but none of the performance graphs show any problems at all.21:16
h00kthat explains it pretty well.  At first it's counterintuitive, but it does make sense21:16
bigbashthe opterons we are looking at are 12-core, and there will be 2 procs per host21:16
h00kI mean, with something your scale, you may not have issues running each VM with 2 cores21:18
h00kbut if you may grow, know it can/will actually slow things down if they're waiting for 2 cores to be open at the same time21:18
bigbashgood to know21:19
h00kbigbash: I'ma cruise and study, let me know if you have anything I can try to help with21:34
bigbashI'm sure I can think of more things :)21:34
bigbashSo much to learn and play with21:35
EgyParadoxbigbash: Thank you for notifying me22:25
bigbashEgyParadox, sorry I was too late notifying you22:26
EgyParadoxno its ok :) I didnt know the new date was today22:27

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!