Solved

Ncomputing & Vmware cpu sizing

Posted on 2010-11-28
4
1,966 Views
Last Modified: 2013-11-06
Hi,

I planning to implement 2000 ncomputing devices, for one of my customer.
Ncomputing is a sort of thin client, which as no CPU, no disk no memory, it works the same way as Terminal server.
It's a terminal access device, which allow you to share a single PC or server. You can have up to 30 access devices sharing the Same Server OS.
My plan is to use vmware ESXi, and have mutliple virtual machine and connect 15 clients using the ncomputing technologie to eache VM.
The recomandation for  a single physical machine is 4 gig ram and quad core processor, and I'm struggling to find out what will be the best recommendation with Vmware Esxi for the CPU.
Will I need one Vcpu  per machine ? 4 Vcpu ? and each Vcpu should be assigned to a core ?
Knowing that they will be using Office 2007 or 2010 , Antivirus and Microsoft Steadystate.
Basically, I need to know how much vCPU i will need to have and how many physical core I will need to use per VM in order to get reasonable performance for 15 users using the same VM?

Thanks
0
Comment
Question by:ynhari
  • 2
4 Comments
 
LVL 28

Expert Comment

by:bgoering
ID: 34225875
You are basically going for multiple virtual desktops - as such you could probably get an idea from the VMware View sizing guide (http://vmware.com/files/pdf/Mainline-Virtual-Desktop-Sizing-Guide-VMware-View-vSphere-WP-EN.pdf)

I looked at NComputing (I think anyway) at VMworld this year, and they claim to be a bit more efficient on the ESXi requirements than View - if you can quantify that number you could add that many more desktops to the sizing guide.

For example, the reference architectue for the paper was a blade server with 2 Xeon 5570 2.93 GHz Quad Core processors and 96 GB of ram. It would support 200 users over PCoIP. If your NComputing would support 50% more users on the same hardware then you could expect 300 users per blade with a similar configuration. If it supports 25% more users expect 250 users per blade.

As far as the VM configuration - with the Xeon 5570 it supports hyperthreading (as does 56xx processors) so a blade with 8 cores yields 16 logical processors. VMware likes the hyperthreading in Nahalim and Westmere processors. With VMware you can overcommit processor and memory so to support the theoretical 300 users per box (or blade) at 15 users per vm - you will need 20 vms running. with the 16 logical processors you would have no trouble allocating 2 processors per vm, and likely not have any trouble allocating 4 processors per vm. I would recommend starting with 2, and incread to 4 if necessary. Your 4 GB of memory per vm sounds reasonable.

Hope this helps.
0
 

Author Comment

by:ynhari
ID: 34225946
Thanks for those information, yes it helps.

Just need some clarification, you mentioned that with 16 logical processors I will have no trouble allocating 2 processors per Vm, does it mean that the maximum VM I can have will be 8 on the physical server?

Or you mean, I will have 40 vCPu's sharing those 16 logical processors, which is 2 or 3 vCPU per cores ?

0
 
LVL 28

Accepted Solution

by:
bgoering earned 500 total points
ID: 34226050
The latter - 2 or 3 vCPU per core is reasonable overcommittment of processor resources for most workloads. You indicated your clients will mostly be using MS Office type applications that aren't alll that cpu intensive. I typically will use up from 6 up to 10 vCPU per core in server workloads depending on how cpu intensive the server workloads are. Generally I run out of physical RAM well before encountering a CPU bottleneck.  

From http://www.vmware.com/pdf/Perf_Best_Practices_vSphere4.1.pdf:

"In most environments ESX allows significant levels of CPU overcommitment (that is, running more
vCPUs on a host than the total number of physical CPUs in that host) without impacting virtual machine
performance.
If an ESX host becomes CPU saturated (that is, the virtual machines and other loads on the host demand
all the CPU resources the host has), latency-sensitive workloads might not perform well. In this case you
might want to reduce the CPU load, for example by powering off some virtual machines or migrating
them to a different host (or allowing DRS to migrate them automatically)."

That is an excellent document worth a read before implementing your vSphere environment.

Good Luck
0
 

Expert Comment

by:sahamh512
ID: 34226939
YNHARI -

I am the VP/GM of the Americas for NComputing

I want to make sure you are talking to us and we are answering your questions, etc

I would be happy to put you in touch with one of our Senior System Engineers

Let me know

Steve Halland
shalland@ncomputing.com
0

Featured Post

Comprehensive Backup Solutions for Microsoft

Acronis protects the complete Microsoft technology stack: Windows Server, Windows PC, laptop and Surface data; Microsoft business applications; Microsoft Hyper-V; Azure VMs; Microsoft Windows Server 2016; Microsoft Exchange 2016 and SQL Server 2016.

Join & Write a Comment

Last article we focus in how to VMware: How to create and use VMs TAGs – Part 1 so before follow this article and perform the next tasks, you should read the first article how to create the TAG before using them in Veeam Backup Jobs.
In this article, I will show you HOW TO: Perform a Physical to Virtual (P2V) Conversion the easy way from a computer backup (image).
Teach the user how to delpoy the vCenter Server Appliance and how to configure its network settings Deploy OVF: Open VM console and configure networking:
How to install and configure Citrix XenApp 6.5 - Part 1. In this video tutorial we have explained step by step installation of Citrix XenApp 6.5 Server on Windows Server 2008 R2 is explained in this video. We have explained the difference between…

757 members asked questions and received personalized solutions in the past 7 days.

Join the community of 500,000 technology professionals and ask your questions.

Join & Ask a Question

Need Help in Real-Time?

Connect with top rated Experts

22 Experts available now in Live!

Get 1:1 Help Now