Cloud Storage Stability

Every week or so one major internet service or another goes down for a moment, Amazon S3, Google Apps, Twitter etc... As a result I get a few emails using this as proof that the cloud is not reliable for business storage. While I agree that various providers have to improve their uptime, this does not mean that the whole cloud needs to be thrown out.

Let's face it, if you store data in the cloud there are a hundred variables between you and your data and if any one of those variables decides to, well, be variable, then you may not be able to get to your data for a period of time. This does not mean that you can't use the cloud, it means that you can't put data that you are going to need immediate access to solely in the cloud. 

What this does mean is using a hybrid model for cloud storage. As we demonstrate in our latest video "What is Hybrid Cloud Storage?" a hybrid cloud is an appliance that is placed on the customer's site to act as a intermediary storage location for data that is in route to the cloud. The appliance serves many purposes: translation from CIFS/NFS to more internet friendly protocols, local cache for rapid restores of last copy of a backup or archive and as a place to get to data that would otherwise be inaccessible due to some sort of connection issue. 

The appliance can be sized typically to store as much data locally as you feel comfortable with, while simultaneously replicating that data to the cloud storage provider. We see small business versions of this today from companies like Dropbox, Soonr and others, which have local clients or agents that you load on your desktops/laptops making your personal system a hybrid appliance. These agents synchronize or archive data to their cloud storage. To some extent the cloud extensions that I wrote about in a previous entry makes the backup or archive application that hybrid appliance on your desktop. 

As you expand, installing and managing clients becomes unwieldy and the need for a centralized appliance become obvious. Companies like Iron Mountain, Nirvanix, Axcient and others are well down the road to providing this sort of intelligence. 

Typically,  you can decide how much storage you want local with these systems, as well as how many copies you want sent to the cloud and how quickly. This gives you immediate and internet independent access to your data while at the same time you still the protection of having that data replicated to the cloud. It is not the amount of data you are storing locally that drives up cost, it is the management of that data. When acting as a cache, the appliance burdens you with little if any additional management requirements.

The next questions are: is the cloud a viable final resting place for your data and what are the cloud service providers processes for protecting your assets when they have them? Are we to the point where we need an independent third party to audit and backup our investment in the cloud?

We'll try to address these thorny issues in our next few entries...