Server upgrade

I had planned to start a server upgrade this holiday - however due to the NTLDR is missing error message appearing during a server reboot, the upgrade was rushed forward to this afternoon.

After attempting a repair from the Server 2003 install CD (which failed), I decided to swap out the 120GB main drive for another I had lying around. Then whilst the fileserver was re-installing Server 2003, I put the dead drive into an external enclosure and attached it to the new server. Luckily the drive was still alive and I was able to copy the 50GB+ of Virtual Machines to the new server.

To cut a long story short, everything is up and running now, with my setup split across 4 servers now instead of the previous 3. You'll notice a slight speed increase from the upgrade - as the slickhouse homepage now takes around 0.5 seconds to generate, rather than the previous 0.7. I'll look into moving MS SQL Server to a new virtual server, leaving MySQL on its own to occupy the full allocated RAM, in the near future - which may reduce that time even further. The new setup is as follows:

Router:

• Smoothwall O/S
• Mini-itx EPIA 5000 board (~550MHz VIA CPU)
• 512MB RAM
• 40GB HDD

Backupserver (not always on):

• Server 2003 O/S
• AMD Duron 950MHz CPU
• 1GB RAM
• 1 x 80GB O/S HDD
• 4 x 250GB Data HDD

Fileserver:

• Server 2003 O/S
• AMD Athlon XP 2GHz
• 2GB DDR RAM
• 1 x 120GB O/S HDD
• 4 x 250GB Data HDD

Virtualserver:

• Server 2003 O/S
• AMD AM2 4000+ 2GHz (Dual core)
• 6GB DDR RAM (max of 16GB)
• 1 x 250GB O/S and Virtual HDD

The Virtualserver itself still uses Microsoft's Virtual Server 2005, though I tried installing VMWare's ESXi, but it doesn't support the motherboard's onboard SATA controller (AMD 780G Southbridge). The point to mention is that it can support up to 16GB RAM, using 4 x 4GB modules - however 4GB DDR2 modules are still a rarity. At the moment it's using about 1/2 of the 6GB RAM running 7 virtual machines (all Server 2003 at 256/512 RAM allocation).

I've decided to stick to Virtual Server 2005 for the forseeable future, as it's very quick to get a Virtual Machine back up and running if I encounter another problem like today's. It's simply a case of copying across the vhd and either creating a new machine (pointing to said vhd), or also copying across the vmc file to keep the existing machine's settings.

Separating the 1TB storage from the virtual servers should make storage upgrades easier in the future and also allow for less downtime when carrying out maintenance. Let's see?