Thursday, December 29, 2005 VMware ESX Server performance comparison on AMD and Intel servers picks up the Tom's Hardware ESX server analysis on a variety of server platforms. Page 17 has the overall results but the entire thing is well worth the read as Alessandro points out. Our server platform is the Dell PE 2850 running hyper-threaded 3.4 Ghz processors with 8GB of RAM. It's SAN attached with a variety of high I/O virtual guests including Exchange 2003 and SQL 2000.

Wednesday, December 21, 2005 VMware Player 1.0.1 released to Address Security Issue

We all needed to know about this one. Spread the word...

Tuesday, December 20, 2005

VMTN Blog Reports Some Interesting Snippets

Running VM Workstation as a service may not be supported, but the concept is truly interesting. Definitely not production stuff but the flexibility is nice for technical users like myself.

VMware not based on Red Hat - official

Some suggestion that VMware's ESX was based on Red Hat proves untrue.

Tuesday, December 13, 2005 VMware launches Community Virtual Machines

Great news! As Alessandro points out, now we can share VM's on a global scale. Given licensing costs for some operating systems, my guess is this list will remain pretty much open source unless we see a platform vendor providing some freely licensed VM's.

Friday, December 09, 2005

Microsoft grooms Virtual Server to be a true VMware challenger | InfoWorld | By Oliver Rist

Microsoft is truly entering the game and upping the stakes with their latest Virtual Server 2005 R2 offering. It has some interesting features which show some maturity and advantage. What we need to see now is real world scenarios and performance.

Thursday, December 08, 2005 Whitepaper: VMware ESX Server Performance and Resource Management for CPU-Intensive Workloads

Helpful whitepaper available.

To MSDE or Not MSDE...That Is The Question

Within the last several weeks, I've taken action to address a problem with Virtual Center 1.3. When I upgraded from VC 1.2 to 1.3, I aggravated an issue regarding the performance graphs. For a while now, I've been experiencing gaps in graphed data. With the upgrade, it became a real issue (see picture).

Working with our support vendor, they determined that running MSDE as our SQL backend must be the problem. However, I remain unconvinced.

First, there is much debate about whether this configuration is supported or not (see discussion threads here and here) I maintain that MSDE is SQL lite and so I am running SQL for Virtual Center. I know about the limitations but I'm nowhere near meeting those. I don't think you should be required to spend money on a full-blown SQL license just to run Virtual Center. Products from Cisco's Security Agent Management Center to NetIQ management products use MSDE as their database.

Second, there was no exploration of the database content itself to see if it was "corrupt" as the support engineer claimed based on the graphs. More investigation would need to be made before identifying MSDE as the cause.

Lastly, we made another change which addressed the issue. We changed the number of samples taken and time interval. This seemed to do the trick. Changing the time interval from 5 minutes to 10 minutes and samples from 288 to 144 fixed the graphs. I don't believe I should have to do this and I already have a gripe about the lack of real-time averages to gauge real performance but it's resolved for now.

Is it MSDE? I won't know without a more thorough investigation if it happens again.