I was finally able to get a trial copy of ESX. Or rather, a co-worker was. I think my spam filter had it in for me and VMWare. The install of ESX was pretty straight forward, or so our resident Linux guru told me. All I know is that he got it running in a few minutes, and then it was time to get something loaded.

My prior experience with VMWare Converter was decidedly mixed at best. A basic 2003 server finished the conversion fine, and worked without trouble in the Server product. However, another server, our main information/application server, while not indicating any troubles during the conversion, failed to load in Server. All I got was a black screen. It would not even load the VMWare BIOS screen. I did some quick searching, but the only thing I could find online was about an incompatibility with a 2000 SCSI driver. My system had the suspect file, so I decided to leave it be and move on. My third attempt at a conversion was our Great Plains accounting server. This time, the conversion process failed.

Conveniently, the ESX server was now up and running, so just for the heck of it, I pointed the converter at the image made of our App server, the one that would not start up in VMWare Server. The tool took some time, about 45 minutes (the image was over 100GB), but lo and behold, upon finishing and booting, the server was running just fine inside ESX. I have no idea why it would not work in VMWare Server, perhaps it was too large a file. At any rate, with my new found confidence, I decided to give our Great Plains server another go. I removed some unnecessary databases/backups off the server, bringing down the file size, and decreasing the time requirement. Progress was smooth and reasonably quick, and once again, ESX started the OS sans problems.

I am running some additional benchmarks. So far, they point a very positive picture for ESX, but they are only preliminary. It is not a true apples to apples comparison, as the hardware running both servers is older and thus slower than the server running ESX. However, so far, ESX is putting up some impressive results. Running a single VM, with 2 cores assigned to the VM, Sandra is showing numbers on par with the live system. Giving the VM 2 additional cores shows a tremendous speed increase. While you may say it is not fair to compare apples to oranges, to us, it makes sense. The newer server is certainly not twice as powerful as the live servers running our systems. But if it is capable of consolidating our two most demanding servers into one, without sacrificing speed, it is a big win in manageability for us.

Tomorrow, I should get some results of both VMs running simultaneously, using Sandra as well as SQLIOSIM. At this point in time, the experience has been running local disks. Of course, VMWare’s advantages shine more in the SAN world. If we were not purchasing a SAN in the near term, I may not be as excited as I am. I just hope the speed of the SAN does not disappoint compared to the local disk array I am currently using.

Bookmark to: