The first ACM symposium on Cloud Computing, which will bring together a range of experts on the topic, is going to take place this June, and we noticed an interesting proposal for a new kind of cloud storage proxy. The basic idea is a good one, but to some extent it falls into a familiar trap: the tendency to look at cloud computing and cloud storage through a similar lens.
The new paper proposes a kind of extension of the RAID architecture we mentioned last week. The researchers are calling it RACS, or Redundant Array of Cloud Storage, and they describe it as a cloud storage proxy that actively picks the least expensive provider whenever it sends data to the cloud. Data would be striped across multiple cloud storage providers.
This notion of price arbitrage has also been springing up in the cloud compute world. In that space, it can make sense – you pick whichever platform has the lowest cost for compute cycles at any point in time and send your jobs to that vendor. This should work well if the processing and the data that needs to go with the compute is small enough.
For storage, this approach, and the RACS idea, could also prove effective with small data sets. Individuals or small companies could benefit if they are looking to move old data to the cloud in the cheapest way possible. But once you start dealing with primary data, and move up to the scale of hundreds of gigabytes or terabytes, this arbitrage model would fail.
If your system were constantly picking different providers based on cost, shuttling one dataset here and another there, you would be left with far too many failure points. Your files would be scattered across the cloud and moving that data around would be difficult. Performance would suffer - files called back from different clouds would be reinstantiated at different speeds, creating headaches for your users. The results would be inconsistent.
There is a simpler way – we designed the Filer to solve many of the same problems RACS looks to address. The recently launched full version of the Filer allows customers to choose from multiple storage providers, or use several at once. We think our customers should know where their data is, and the Filer gives them that control. Customers can store data on two different cloud volumes at once. (We are also considering adding cloud mirroring capabilities to a future release.) Plus, we simplify everything, so our customers don’t have to worry about the fact that the cloud providers have different APIs, layouts, billing procedures, and more. Nasuni takes care of all that.
The RACS paper notes the trouble with vendor lock-in, which can make moving from one provider to another prohibitively expensive, but our technology eliminates this problem. For Nasuni customers, moving from one cloud to another is just a matter of drag and drop.
Finally, Nasuni levels the cost of cloud storage – our customers only pay for what they use. Any questions? Send us a note at feedback@nasuni.com.