I call it the "common sense" methodology. Use it when it makes sense. Every environment and every situation is different. Each client has a different objective and goal when it comes to technology (and of course that changes over time too!). If the technology makes sense, use it. If not, then move on. It's just like Metaframe. It's a great product, but it's not a cure all. There are some situations where it not a suitable choice and another solutions needs be devised. Joe _____ From: thin-bounce@xxxxxxxxxxxxx [mailto:thin-bounce@xxxxxxxxxxxxx] On Behalf Of Ron Oglesby Sent: Wednesday, December 08, 2004 8:44 PM To: thin@xxxxxxxxxxxxx Subject: [THIN] Re: VMWare ESX Server I would totally agree. Obviously there are tons of servers/apps that are "not a perfect fit". But on the consolidation side, servers that are really under utilized (<30% on any of the core four resources) it's a no brainer. Now if you have servers that DON'T perform as well in the VMworld given the same amount of memory and other resources, then you have to make the business decision if the benfits of a VM are wroth the cost of making TWO or more VMs to split the load. On a consolidation front two studies I have done for clients put the VM to Phys machine (when looking at consolidation) at about a 1:3 cost ratio. On the non consolidation front, those non0strike zone servers takes that up to a 1:1 or 1:1.25. Now if it's a mixed environment of strike zone and non-strike zone consolidation servers then we are somewhere in between. Anyway if you are going to POUND a server then maybe physical is better unless you need one of the other features, like near instant recovery. Then the additional cost becomes part of your DR scheme and not really provisioning itself. Ron Oglesby Director of Technical Architecture RapidApp, Chicago Office: 312 372 7188 Mobile: 815 325 7618 email: roglesby@xxxxxxxxxxxx _____ From: thin-bounce@xxxxxxxxxxxxx [mailto:thin-bounce@xxxxxxxxxxxxx] On Behalf Of Chris Fraser Sent: Wednesday, December 08, 2004 12:47 PM To: thin@xxxxxxxxxxxxx Subject: [THIN] Re: VMWare ESX Server Hi there Brian I would disagree that virtual anything can never be as good as the real thing. In fact I think VMware ESX-based anything can be quite a bit better than the real thing _BUT_ you need to have a good understanding of your workload and usage patterns (sound familiar anyone?). Hardware-cost and form-factor are trivial expenses compared to the overall TCO or $$$ impact of downtime. I've seen scenarios where the vendor, with a minimum of troubleshooting or problem isolation, pointed the finger at VMware. After further investigation we were able to determine that the issue was actually the application or the OS and it was only exposed or magnified by running on VMware. VMware will not give you something for nothing. It is absolutely key that you understand what you have and where you need to get to and then plan and build it accordingly. _____ From: thin-bounce@xxxxxxxxxxxxx [mailto:thin-bounce@xxxxxxxxxxxxx] On Behalf Of RMC - Brian Hill Sent: Wednesday, December 08, 2004 10:24 AM To: 'thin@xxxxxxxxxxxxx' Subject: [THIN] Re: VMWare ESX Server I've never been a fan of VMs in the production environment. I have used them for development a few times, but even then, I prefer a "real world" scenario for development as well. I've had DC's that were flaky, with the only variable being VMWare. Dunno, maybe it's just me. Hardware cost are relatively cheap these days, the form-factor is small, so I don't think VMs buy you a whole lot. They have their place I suppose, but they are not something I pursue. Virtual "anything" can never be as good as the real thing IMO. Brian -----Original Message----- From: Jennifer Hooper [mailto:jennifer.hooper@xxxxxxxxxxxxx] Sent: Wednesday, December 08, 2004 12:23 PM To: 'thin@xxxxxxxxxxxxx' Subject: [THIN] VMWare ESX Server Hi Guys - Here's the situation. We are about 1/4th of the way through converting our production data center (full of old, out of warranty Compaq servers) to IBM BladeCenters running VMWare ESX Server 2.0 (I think... whatever the latest version is). We use Platespin to convert the physical server to the VMWare image and move it over to the Blade it's going to live on. Right now, we have an average 5 servers per blade planned, and several have already moved over, because the hardware failed that they were on. However, some of the application folks are uncomfortable with this solution, not so much the Blade technology, as the VM technology. Needless to say, we're already experiencing failures, and stuff not running right - performance issues, network issues, etc. (Can you believe that they are going to run our Root Domain Controllers on this?) I have already experienced a drag on one of my Citrix servers that moved to virtual space, and can't fix it up. So what I would like to do before things get too much more hairy, is to try to find out what the success rate of running something like this in production, and if there are a lot of people out there doing this. Feel free to share with me any nightmare stories too! :) Thanks much! Jen Jennifer Hooper Peregrine Systems, Inc. Sr. Network Engineer mailto:jennifer.hooper@xxxxxxxxxxxxx