In one of our clients implementations of customized Data warehouse / integration , we have kind of reached a level where every half an hour there is some or the other ETL running and populating the Database. Also its shared with other teams and hence we have other ETLs as well running on this box.
As per my experience I think its time to buy another server exclusively for our purposes , but i will like to know from you gurus that what actually decides to go for another server , what reasons are valid to justify this purchase etc.. Any other options feedbacks or customer stories that we have for the same situation !
Thanks for you inputs !
*Note - Title made more descriptive - Content Editor*
So what actually made you decide you need multiple servers? Was it the data frequency , volume etc , how did you approach to get more? Any feedbacks , pointers?
kduke wrote:We have multiple servers. The ODS is considered critical so it has its own server now. I wish we had a grid instead.
Data volume. The other projects could overwelm the server when they kicked off. ODS was more critical. EDW jobs were batch and could be restarted easily. ODS is near real time and needed to run continuously.
I still think a grid is more flexible and adaptable to either real time or batch.
Most of my recent sites have been blade-type servers. We just pop in another blade to add CPU / memory.
Most of my sites have dedicated DataStage servers. Once the other applications realize what a resource hog it is, they quickly depart for better servers!