It is always a challenging task to benchmark your system without using real data. I am in the midst of a project where we would like to scale out, i will be migrating from pretty big HP DL 580 Gen 8 ( 1 TB RAM, 120 Cores with hyper threading enabled) with backend as FUSION IO storage. One of my goals is to make sure i don’t take a hit on performance when i scale out. I will be testing different storage systems along with different branded servers. I cannot take my actual production data as it is highly confidential. I will be using Hammerdb with Tpc-H standard since our environment has 80% read and 20 % write. Though i can’t use production data,i still want to get as close as production like data access pattern in my test environment. There are many different ways to tackle this, my plan is to capture few basic metrics from my actual production environment and try to push my test environment to the same limits. For those who haven’t used hammerdb for SQL Server before please refer to Kendra’s tutorial on Hammerdb, she has it for Tpc-C but same steps will work for Tpc_H.