The servers are grouped so they share workloads and ensure computing is completed if a single server crashes. Will Big Blue find the concept notoriously tricky as others have? Stephen Shankland worked ...