Virtualizing OCS 2007 Speech Server
After talking with my good friend AC and reading a recent post he wrote on making the switch to VMWare I decided to experiment with setting up some virtual environments. You have to realize that I have never been a fan of virtual environments. I come from the old school (where we only had 640K of memory) and I don't like giving up any CPU cycles unless I have to. Also in my previous job we just stood up more servers when ever I needed a new environment for something.
I write a lot of experimental Speech Server applications just to test something or learn some new aspect of the tools. These tend to clutter up my hard drive mixing these experimental apps in with the real apps.
I've been toying lately with the idea of standing up a production environment using a virtual so after reading Andrew's post I decided to just do it. I respect Andrew's opinions very much (even though he is a SharePoint guy) so I downloaded and installed VMware Workstation.
Installation went very well and was fairly straight forward. The interface is very clean and easy to use. I had no problems creating a virtual machine - just some mouse clicks and then set back and wait.
Once you have a virtual machine you have to install an OS and there is where I ran into a problem. The normal procedure is to put a bootable OS CD in the drive and then start the virtual. My MSDN CDs are not bootable so the virtual wouldn't recognize it. What to do? A quick trip to the VMware web site and a few searches there I learned that you can point the virtual's CD drive to an ISO image on your hard drive and it would install the OS from that. A quick copy of the OS ISO image fro the MSDN CD and I was off and running. After finishing I simply pointed the drive back to the proper place.
I now had a base Windows 2003 Server image that I wanted to preserve so the next step was to clone that image which again was just a few mouse clicks and set back to wait.
I now had a base 2003 image and a clone which I called "OCS 2007 Speech Server Base". This is the virtual that i installed Speech Server on. After that was done and the install was tested I cloned this virtual also so that I would have a working production environment. I did this so that I can quickly and easily clone as many OCS Speech Server environments as I need.
So how does it work? Well, after a little tweaking it works very well. Using a soft phone on another machine I placed a call to the virtual and the WelcomeToOcs application answered but the sound quality really sucked. It was broken up and very choppy sounding. This was a disappointment and at first made me wonder if this concept was going to work. But rather than give up I started do some digging and found that the default memory allocation for a new virtual appears to be 384 MB. I quickly changed the memory to 1 GB and tested again. It sounded sweet. But this was to be expected as Speech Server loves memory.
All of this only took about 3 hours from downloading VMware Workstation to having a working Speech Server environment. I'm very happy and plan on pursuing this some more. As I do I'll blog and keep you updated.
Points to remember:
- OCS 2007 Speech Server works well in VMWare
- You need to allocate at least 1 GB of memory
- You should create both a base OS install as well as a base Speech Server install.
- Always clone the base installs and work off of the clones not the base installs
I would love to hear about your experiences with virtual machines - especially if you are using Speech Server.