Using an object storage system as a traditional array alternativeDate: Sep 30, 2013
According to Phil Goodwin, principal architect in Cognizant's IT Infrastructure Services Group, an object storage system is a viable alternative to a traditional shared storage system because of its scalability and flexibility. In this video from the Storage Decisions seminar in Chicago, Goodwin sat down with Rich Castagna, editorial director of the Storage Media Group to explain why alternatives to shared storage are often sought out, and why object storage systems can help.
We're hearing a lot lately about alternatives to traditional shared storage systems. But what's the problem with those systems? Where are the current products failing?
Phil Goodwin: Well, there are really two limitations of current architectures. One is recoverability. We're simply getting such a massive scale of storage that it's difficult for organizations to do things like recover from simple errors -- like a media error or something of that nature -- or even entire systems. Contemplate having to recover tens of petabytes of information from one location to another location. It's becoming very difficult.
The other one is inflexibility. As we've seen more and more implementations of virtualization, we have some wonderful capabilities to move applications from location to location, but it's all limited by your ability to stage data. In fact, that's really one of the limiting factors that you have, both in disaster recovery as well as in simply having a flexible architecture to put data and applications where they really belong at any given point in time.
Object-based storage is often cited for its scalability. How does it differ from file or block systems that are commonly used today?
Goodwin: Well, the key difference between an object storage system and other more traditional architectures is that object storage really separates the what from the where. It's really trying to solve that flexibility problem that we were talking about in virtual infrastructure environments, and move it more toward the ability to move data and to place data independently of the underlying storage infrastructure.
So, instead of having a traditional RAID environment, for example, or LUNs [logical unit numbers] and volumes, what you have are hashing algorithms that actually place the data where it ultimately will reside. It gives that much greater flexibility to the application and to the data management software, and illuminates the problem of RAID, RAID-DP, double-parity RAID, RAID 6, things like that in trying to recover from different kinds of scenarios.
Can object-based storage be used interchangeably with file or block storage?
Goodwin: Not really, because you still have to have an application that will be written to the [application programming interface], like the REST API or the OpenStack API or something like that that can address object-based storage. So, there are certain abstraction layers that you can put in there that abstract the application from what the data's doing, but there does have to be that layer in there that can do the translation for the application to say, 'OK, instead of going to this LUN structure, I'm going to go off to this object-based structure.'
About the expert:
Phil Goodwin is a senior manager and principal architect in Cognizant's IT Infrastructure Services group, where he assists clients in the development of adaptive storage architectures, storage management best practices, backup and recovery, disaster recovery, and data archiving.