Cisco: Plan For Failure In Big Data Backup

When it comes to backing up massive amounts of data, Cisco Big Data consultant Robert Novak's message might as well be "expect the worst." That may sound pessimistic to those inexperienced in handling big data, but considering the number of things that can possibly go wrong, it's a smart mindset.

"Today, if somebody asks, 'How many servers do I need? How much data do I need?' Well, being in Cisco sales, a million servers would be really nice," Novak said.

Joking aside, Novak said it's important to consider what heating and cooling servers will need, and what purchase orders the CFO will actually put in, but err on the side of caution.

"You're going to want to plan for failure," he said.

"Plan for at least 1 percent [of servers] to fail a week, or at least one server if you don't have a hundred in a cluster. You're probably going to be set up to at least sustain your environment," Novak recommended.

That could be a tall order to submit to managers or financial officials, but consider what Novak had to say on the point of cost:

"More servers may cost less than data loss. It may be better for your career as well," he said.

PUBLISHED JUNE 23, 2015