After last years complete rebuild of the network cabling infrastructure and data switches, the major project for this year is to rebuild our domain from scratch. We have decided to do this using a HyperV environment. The reasons for doing this are numerous but the main two being cost savings from going from 24 physical servers down to 9 virtual servers under the current draft design and the future ease of management and ability to add new servers without the cost of purchasing new hardware.

So the point of this post is to show you our 1st draft of the server infrastructure we came up with and if you feel the need to comment then please do so.

Current Infrastructure

We currently have the following physical servers

The HyperV server currently runs 5 servers

1st Draft

Here is an image of our 1st draft of the new infrastructure

The thinking behind this is quite simple.

• Anything connected to the outside world should be separate from your internal network so the two proxies, one for outgoing traffic and one for incoming traffic are kept as physical servers
• Our Sun Ray thin client system uses Sun Solaris servers to deal with the clients and point them in the direction of the Windows TS load balancer. They are very specialised and not ideal for HyperV
• Personally I think you should have at least 1 DC as a physical server, you can then have redundant ones in the HyperV environment
• Our Veritas backup server uses a direct attached tape carousel which HyperV does not recognise

We will then purchase 3 new servers for the HyperV cluster backed up by a SAN for the storage of the HyperV data and VHD’s

So that's the start of the process, if you have any comments then please put them down and maybe we can start a discussion around HyperV