Problem: Having as much data as possible can help more accurately predict outcomes, but that benefit means having to keep track of a lot more information. Large volumes of data can easily translate into maintenance headaches for architects. Dr. Satya Ramaswamy of TCS said data maintenance problems can be particularly evident in scenarios such as switching platforms, and an OS level upgrade to all nodes in a cluster.
Solution: Think ahead of time! Data maintenance risks can be reduced by using multiple clusters, said Ramaswamy. "It would also be worthwhile to plan for a big data archival solution."