Today: 2 hardware servers - one is a sql server, the other is the rest of MOSS SharePoint 2007 sp1. The hardware is 32 bit, running Windows Server 2003 R2. 4 GB. Sizing: We are importing about 4600 "user profile" entries - however, there are not that many "users". The import is bringing in sa accounts, some groups, etc. We have approximately 1700 accounts corresponding to individuals, and the rest, apparently are roles and other things. We have over 1500 SharePoint sites - 2 content databases, about 2 dozen site collections The plan of which we are trying to conceive Having never installed SharePoint before, we are doing lots of googling, reading, etc. The idea we are thinking is to virtualize the "sharePoint" machine. In doing so, we want to replace the current old machine with one machine for web front end type work and move some of the sharepoint processes that are cpu intensive and relating to crawling/indexing over to a second machine. Does that make sense and would it be doable? Right now, the plan is for the additional machines to be configured as 32 bit, because we don't want to deal with the possibility of moving the several thousand SharePoint sites to a new bit platform and potentially breaking things. By keeping things as much the same as possible, then the moves should be relatively painless - we hope. I know that the 64 bit decision is not a traditional decision, but it seems like the safest way to go at this time. The future is going to force the 64 bit decision at a later time, and that will have its own consequences. To go this route, we are talking about adding the two new machines to the farm as defined by the first server. I presume we would have these new machines pointing to the same sql server databases (since otherwise, it would be a different farm, rather than new machines on the new farm). One avenue we are seeking is to place these behind an F5 load balancer, then to slowly move users from the old system to the new system, allowing us to assure ourselves that the vm configuration (which we would parallel the existing machine) can handle the load. We need to figure out how to configure the second machine to do just index and crawling (unless there are other non-interactive type processes that would also be beneficial to move over). Once it is up and running, then we would point the old machine at the new index/crawling machine and stop running those processes on the hardware. Has anyone else gone through these same types of decisions and directions? Are there other issues that we need to be certain to consider during this process? We really appreciate insights you might have from your own experiences.