What is your consolidation ratio after virtualizing?
We are still in the process but have gone from roughly 85 physicals servers to four ESX hosts, but added an additional 20+ VM's in the past year with 19 physical servers remaining.
We went from 76 physical to around 20 physical, the 4 ESX servers have about 55 VM's on them.
We are still consolidating (Plus doing a VI3 Migration at the same time). 600 servers have been virtualized (Out of a 1000). We expect an 80% consolidation (Hoping for 90%). Number of Hosts=80 with room to spare. Virtualizing our Citrix environment has been challenging at best.
So far in our main site we've gotten 190 servers onto 14 hosts with PLENTY of room to spare and work still to be done. I believe we should be able to get another 70-100 guests on our existing infrastructure
Or total environment right now is 230 over 22 hosts, but that looks a little skewed because we started deploying VMware in small remote offices because it helps us separate duties (we used to stack server apps/roles on one machine), and gives us more hardware independence when hardware is out of warranty.
I'll have a better idea towards the end of the year.
6 ESX hosts and 78 VM's, 16 cores, so about 5 per core.
However at least half of the VM's were born as VM's, but I'm still counting it, as it is still a "consolidation ratio" since it would be a 1U or 2U anyway. There are still at least 20 or so systems that can be p2v'd but management is going all Cover Their Ass on me and won't let me do it because vendors say "no". FUUUUUD
My hosts are non Core Xeon's, 4 x 2cpu IBM Blades and 2x4cpu IBM x366's. All of the hosts were purchased just before the Core processors were released.
I've asked for $100,000 in '08 to replace them all with dual or quad cores - we'll see how that pans out
We had about 60+ servers and we moved almost everything ( 95+ %) to a VM across 4 ESX hosts. The only physicals left were an exchange server and a few other Windows boxes (we couldn't convince the corporate IT group to go virtual for these ).
Like what I'm reading so far. Virtualization has certainly made \*my* life easier from a hardware support stand point.
Have done many implementations before.
Typical is 16:1, have done 1:1 all the way through to 36:1.
Usual config is 4 x dual core AMD, 32GB RAM (DL585 or 6950), 95% on CLARiiON storage, only the special ones on DMX
With a Clariion storage, and 4 very powerfull esx, you cant take easy 5 esx to 100 VMs.
I have three separate farms...
One is an Intel Farm, consisting of DL580 G2s and DL740s. These are single-core 4-processor servers with 28-32 GB of RAM. 16 VMs to each host (or 4 VMs per processor) gives excellent performance...
CPU (%Ready especially) was my bottleneck..
Second farm is an AMD Farm, consisting of DL585 G1s and G2s. These are dual-core 4-processor servers with 32-48 GB of RAM. We're currently running 25-30 VMs per server here... We could probably get more, but memory is the bottleneck here. The G2s have 48 GB of RAM, but the G1s only have 32 GB.
The third farm is in the InternetDMZ and consists of DL385 G1s... These are single-core 2-processor servers with 16 GB of RAM. We get about 10 VMs per server here with great performance...
Im guessing the purpose behind the two different farms (excluding the dmz servers) was the the difference in generation in hardware. Or is there a business case to run amd AND intel within the organization depending on the department or server use?
Glad to see the emphasis and results people are getting from their environments.
I've got 6 hosts (8-way single core), with about 330 VM's running at any one time.
So, that's about 6.875 or almost 7 per core.
Running out of memory though. It always seems to be the bottleneck. (32GB each)
My next 2 boxes are probably going to be some 2-way, quad core, 48GB x3650's.
Im guessing the purpose behind the two different
farms (excluding the dmz servers) was the the
difference in generation in hardware. Or is there a
business case to run amd AND intel within the
organization depending on the department or server
Glad to see the emphasis and results people are
getting from their environments.
No real design decision to have both an Intel and AMD farm... The original farm was a couple of existing reclaimed DL580 G2s... We grew 1-2 host servers at a time... We reclaimed a few more Intel servers and bought some (for VMotion compatibility). We got large enough that I knew I'd have to start a new cluster soon anyway. I wanted to go dual-core AMD for a while, but needed to make a big enough purchase in order to retain N+1 redundancy...
About 9 months ago, I was able to convince the powers-that-be to buy 4 AMD host servers at once (which we promptly filled up of course)
Currently we have 12 Intel host servers and 6 AMD host servers... Within a year, we'll be at 12 and 12 I'm sure, and I'll have to start considering another cluster...
FYI - When I said farm in my original post, I meant cluster.