Hi Guys,

I wonder if anyone has come across a good technical paper/document which describes how to come up with a most efficient design when it comes to creating large filesystems. I just embarked in a new datawarehouse project and would like to drill down more in terms of defining how many volumes groups we have to create, how many LUN's we would required and if there is any rule of thumb that I need to be aware of when it comes to design the underlying storage when creating filesystems specifically for datawarehouses would be great. What would be the appropiate size (volumes groups , LUN's and so on) bearing in mind that the whole system will be 1 TB. We used to just ask the midrange team to create the new filesystem without giving much thought on how the underlying disk is allocate it. I suppose there are so many facts that we need to consider so any hints will be highly appreciated. BTW this datawahouse will be using SAN storage runing on AIX 5.3 64 bits and DB2 8.2 FP12 64 bits. thanks a lot.

cheers.

harby

IBM have something called BCU (balanced configuration unit) for data warehouse. They claim it is an optimal configuration of hardware and software (database and application servers, SAN storage, DB2, etc.) to implement a data warehouse solution. The document describing the BCU in detail is not freely available, but you can request it from your IBM rep.