Search Amazon.com:
Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us $3,400 per month for bandwidth bills alone, and since we don't believe in shoving popup ads to our registered users, we try to make the money back through forum registrations.
«‹›90 »
The Something Awful Forums > Discussion > Serious Hardware / Software Crap > Virtualization Megathread V2: VMs inside VMs
  • Post
  • Reply
Corvettefisher
Sep 8, 2007

GanjamonII posted:

Going back to my situation I had a couple of questions for y'all smart people in here
1) Are the thresholds of eg 10% for CPU ready you're mentioning applicable to larger 8 way VMs? I've seen some blogs online which seem to point to higher CPU ready being acceptable for larger VMs. Eg 40% for 8 core VM. The highest VM I can see right now is sustained at 35% but doesn't seem to be doing any actual work (1-3% CPU utilization).

2) The CPUs in my hosts are 2x 10 core, but with hyper threading shows as 40 logical cores in vsphere. For purposes of calculating over-commitment should I be using 20 or 40 for the physical cores? Eg I have 40 vcpus allocated on one of my hosts, is this 1:1 or 2:1 virtual:physical?

1) I'm not sure what forums that is, but regardless higher CPU ready is more time the VM has to wait to process instructions it hurts performance.

2) You're using 20 physical cores Hyperthreading is not a cpu core, at a high level it work like this. Hyperthreading is not a core replacement, it only helps how the data is processed the the core.

Don't think of it as ratios, think about assigning what the virtual machines need. If it isn't saturating the vCPU(s) it has don't assign more it doesn't make it faster.

# ? May 21, 2013 17:34
  • Quote
Adbot
ADBOT LOVES YOU

# ? May 24, 2013 09:24
  • Quote
Misogynist
Jul 14, 2003

hubthumping

DevNull posted:

So VMware is a service provider now.

http://www.vmware.com/products/data...-hybrid-service
This is honestly a much better use of the Nicira tech than trying to market it as a standalone product right now. Dogfood, etc.

# ? May 21, 2013 19:07
  • Quote
bull3964
Nov 18, 2000

DO YOU HEAR THAT? THAT'S THE SOUND OF ME PATTING MYSELF ON THE BACK.

DevNull posted:

So VMware is a service provider now.

http://www.vmware.com/products/data...-hybrid-service

I'm trying to figure out the pricing here:

http://vcloud.vmware.com/about_services/pricing

Are those unit prices overages? Or are they the cost. It says that 2TB storage is the minimum purchase, but at $0.17/GB/hr that would make 2tb of storage $252,960 /month which is obviously not right.

Having just put our first new vMware cluster in place, we were also looking to supplement our prod environment with a cloud test environment and the ability to seamlessly integrate this in with our current cluster is appealing, but I can't even figure out where things start on pricing here.

# ? May 21, 2013 19:51
  • Quote
demonachizer
Aug 7, 2004

Goon Matchmaker posted:

Known bug if you're on 5.1 with no patches. See http://kb.vmware.com/selfservice/mi...ernalId=2037408

Hmm I am not using windows authentication but it is unpatched so maybe I will try patching and cross my fingers.

# ? May 21, 2013 21:01
  • Quote
Internet Explorer
Jun 1, 2005

OFFICIAL BITCH OF DANBO DAXTER

 People assigning the wrong vCPU amounts for the job drive me up a wall. I work for an MSP and a ton of our environments are set to 8 vCPU on all the VMs. Slowly getting around to changing them.

# ? May 21, 2013 23:50
  • Quote
Moey
Oct 22, 2010

Internet Explorer posted:

People assigning the wrong vCPU amounts for the job drive me up a wall. I work for an MSP and a ton of our environments are set to 8 vCPU on all the VMs. Slowly getting around to changing them.

Ugh. My old boss was so bad with this. Why does my domain controller need 4 cores?

# ? May 21, 2013 23:55
  • Quote
parid
Mar 18, 2004

Internet Explorer posted:

People assigning the wrong vCPU amounts for the job drive me up a wall. I work for an MSP and a ton of our environments are set to 8 vCPU on all the VMs. Slowly getting around to changing them.

I would pay good money for a training video that would walk low level admins and application people through the basics of VM sizing and how its not like what you did for physical machines. I'd use it every day.

# ? May 22, 2013 00:43
  • Quote
hackedaccount
Sep 28, 2009

parid posted:

I would pay good money for a training video that would walk low level admins and application people through the basics of VM sizing and how its not like what you did for physical machines. I'd use it every day.

Does anyone have some good resources for this? I would really like to read more about how to size VMs properly.

# ? May 22, 2013 01:33
  • Quote
Moey
Oct 22, 2010



 This book really does a good job of breaking it down plain and simple.

http://www.barnesandnoble.com/w/cri...n=9781937061982

# ? May 22, 2013 01:42
  • Quote
FISHMANPET
Mar 3, 2007



 So here are some performance graphs for my two VM servers:



This is bad, right? Like terribly colossally bad? By my guess we've got about 200 IOPS to go around and all told, and one of the disks has 300 IOPS of demand (the graph is updated every 20 s

# ? May 22, 2013 16:18
  • Quote
Corvettefisher
Sep 8, 2007



 Yeah that's pretty bad looking storage performance.

# ? May 22, 2013 16:22
  • Quote
evil_bunnY
Apr 2, 2003



 More is better! THREE THOUSAND MILLISECONDS!!

# ? May 22, 2013 16:24
  • Quote
FISHMANPET
Mar 3, 2007



 I'm thinking that more than anything it's the latency that's killing me, right. This stuff all lives in the same rack connected by a single cheap unmanaged switch, I should be hoping for something around 10ms, right?

# ? May 22, 2013 16:24
  • Quote
Corvettefisher
Sep 8, 2007

FISHMANPET posted:

I'm thinking that more than anything it's the latency that's killing me, right. This stuff all lives in the same rack connected by a single cheap unmanaged switch, I should be hoping for something around 10ms, right?

Obviously lower is better. 10-15 is good, maybe on a pretty bad day 20-25ms.

# ? May 22, 2013 16:27
  • Quote
janitorx
May 2, 2002

I'm cuckoo for cocoa cocks!

teh z0rg posted:

This might be a dumb question but is there anything specific I need to do to let an NMS query VCenter Server? anything related to VCenter hardening?

I can poll ESXi in a lab easy... vCenter? Not so much.

I admit I haven't spent a ton of time on this I've been busy wiht higher pri stuff but any help would be appreciated.

Thanks.

Thanks.

A bit late, but if you are on 5.1 and using AD authentication against the API you need to set it to the highest priority provider in SSO. The VI client will still work fine but the API behaves differently.

# ? May 22, 2013 17:30
  • Quote
adorai
Nov 2, 2002

10/27/04 Never forget

FISHMANPET posted:

I'm thinking that more than anything it's the latency that's killing me, right. This stuff all lives in the same rack connected by a single cheap unmanaged switch, I should be hoping for something around 10ms, right?
end users will typically begin to notice slowness once it exceeds 25ms in many applications. they will start screaming between 50 and 100ms.

# ? May 23, 2013 05:11
  • Quote
Adbot
ADBOT LOVES YOU

# ? May 24, 2013 09:24
  • Quote
Corvettefisher
Sep 8, 2007



 https://my.vmware.com/group/vmware/...cloud_suite/5_1

Patch for the Login issue 5.1 U1 caused is now out

# ? May 23, 2013 17:18
  • Quote
  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply
«‹›90 »
The Something Awful Forums > Discussion > Serious Hardware / Software Crap > Virtualization Megathread V2: VMs inside VMs