What Should Cloud Storage Software Do?

Unfortunately cloud storage is often thought of as a big holding tank, where you dump data in case you need it some day, but the truth is that even today you can do much more with it. Assuming that we want the cloud to be more than a data dumping ground somewhere on the internet, what should you expect cloud storage software to do?

As we discussed in my prior entry Cloud Storage Arms Race, cloud storage software can come as exactly that, software that you load on to your servers and storage to create you own cloud or it can be built into the turnkey cloud storage systems that are available. Regardless what this software can do and how it can do it becomes a critical decision point when selecting cloud storage.

One of the first capabilities to look for is dispersion. This is the ability for your cloud software to leverage the fact that it is on a global network and to disperse data throughout that global network. The most obvious use of this capability is as a data protection function. Critical data you may want distributed to four different data centers for example. What makes this important is that some of these software solutions have the ability to dial up or down the number of copies based on business policies.

This can be as cited above to provide an extra layer of protection or it might be to manage the popularity of a file. For example let's say you just released the latest version of your software, you might want to dial up the number of copies and access points for a period of time to meet the initial download demand, then as the initial wave dies down dial the copies down to a primary copy and a DR copy. When the next version of your product is ready for release, you may want to move the prior version of off all the primary access points and just keep one in an archive.

Tied closely into this is the ability for cloud software to be geographically aware. For example you may want to make sure that users on the West Coast download that software from a West Coast data center, users in Europe from a European based data center, etc. Then the software needs to interface with the network to route users to the most geographically close storage point.Billing is another key factor, especially if this software is going to drive an external service. The ability to charge users by the capacity or bandwidth used and then be able to integrate that into an existing billing system is important since eventually you may want to actually make money with this service. In a down economy being profitable is back in style.

From a more infrastructural perspective, things like security and redundancy are a pre-requisite. One capability that is often missed is the software must have the ability to self-migrate or what I call a rolling architecture. We have all migrated one big array to another and I'm sure for most of us it was not a quick and easy process. Multiply that by 10X or 100X the capacity. Cloud platforms have to be flexible, you simply are not going to be able to easily migrate from one system to another. The software will have to support mixed nodes and allow newer technology to integrate seamlessly with older technology. This allows the storage system to age out older components gradually instead of all at once.

A fully enabled cloud storage system is more than a bunch of cheap disk connected to the internet. The software must take advantage of the fact that it is on the internet and make it more compelling than storage in your data center.