The Nirvanix failure

by Robin Harris on Tuesday, 8 October, 2013

Sad to see them fail. But it was never going to be easy to maintain a highly available cloud at much lower scale than Amazon or Microsoft while also paying account teams to work closely with customers to configure the right solution.

In short, all of the costs of an enterprise sale with – at best – Internet margins.

So Nirvanix has failed. But all storage fails. Perhaps not permanently – as did Nirvanix – but whether it is temporary or permanent it fails.

It is the nature of the beast.

The StorageMojo take
Therefore, what we do now with the cloud is no different than what we have done for decades with local storage: we make copies.

The news isn’t all bad. The chances of a local failure and a remote failure being correlated is much lower than the chance of two uncorrelated local failures.

This means that if you use Amazon’s Glacier for deep archive storage it can be your final copy. If it should come down to a legal battle you can demonstrate that you passed the reasonable man test – if you can get the judge to do the math.

More concerned about the data than a legal battle? Copies.

At some point though we have to accept the fact that data loss will always be with us – like conspiracy theorists, perpetual motion machines, taxes and death. Entropy and bugs will see to that.

What’s the line of Niebuhr’s Serenity Prayer?

God, give me grace to accept with serenity the things that cannot be changed, courage to change the things which should be changed, and the Wisdom to distinguish the one from the other.

We must have the wisdom to accept data loss. We don’t have to like it. We should do what we can based on the data’s value to avoid it. But it will happen.

At the media, device, subsystem, system and, yes, at the corporate level.

Courteous comments welcome, of course. Good people at Nirvanix. I trust they’ll land on their feet.

{ 0 comments… add one now }

Leave a Comment

Previous post:

Next post: