Re: Initial configuration questions...

This is a tough one. You're really going to need to focus on what the main use of the machine is going to be. From the sound of it, I suspect that the main use of the machine is going to be the research use. That being the case, you need to think in terms of how to get as much parallelism in the system as possible. That means that you would need to plan on turning on PDQPRIORITY, having an emphasis on virtural memory, large read aheads, etc. Also, for a DSS type system, it's probably better to not strip, but have the chunks be associated with a physical device. The reason for this, is that you will probably want to fragment your tables across your drives to take advantage of parallel scan threads. If the chunks are striped, then the scan thread parallelism can actually cause contention with each other. Basically what you want is for the disk heads to have as little motion as possible. Now for the other side. With OLTP, you probably would not want to have large read-aheads, would want more buffer space. Also, in most oltp applications, parallelism really does not become an issue, and striping is desired. I think that I'd probably go with two instances. This way, the more 'dynamic' activity in the OLTP side would not cause any problems with DSS side. Also, the two systems could be better tuned for their purpose (i.e. read-aheads, virtural memory, PDQPRIORITY, etc) However, I think the most important thing to do is to try to get some consulting onsite. We can't really understand the true nature of your applications via this forum. Madison Pruet