As I was reading the latest debates about theadoption/evolution of iSCSI, I had to laugh (well maybe just grin a bit). I grew up on SCSI, one of many low-level storage protocols that evolved over the years. Not the best one, by far, but the one that made it – kind of like VHS.
The interesting thing is that some folks think iSCSI will somehow make a big difference for customers. The fact is – it won’t. OK, maybe you can now buy all your networking gear from one vendor and reuse some cables you have laying around but little really changes. Not in terms of the BIG cost factors – like how you manage your information.
And yet, the industry treats iSCSI like some kind of revolution. I find it about as exciting as watching a screen saver.
So, what is exciting in storage? SOA. Yes, storage is now all about SOA. If you are in “storage” and aren’t thinking about the implications of SOA and Web 2.0 then you may want to think again. SOA changes how we construct applications by changing how application services interact. The interesting thing is that SOA can be applied to “Information Services” as well.
So, what if, applications “stored” information in a whole new way? Instead of storing bits using tightly-coupled low-level protocols, they would simply interact with an information service. The information service would do all of those little things like storage, protection, security, compliance based on the policies and the whole interaction between the application and information would now be loosely coupled in a Web-service like operation.
Once information is liberated from a “SCSI” attachment to a single application it can then be leveraged by many applications (securely of course). The “value” of the information multiplies. With a common “Service-Oriented Information Infrastructure” (SOII) one can build a single set of information management policies and apply them across hundreds or even thousands of applications. Now that is value!
The fact is, we have been progressing that way for some time. File systems (NAS) gives you more knowledge/capabilities but we can still do much more. Object storage like CAS (Content Addressable) takes it a step further giving applications a greater ability to specify service levels. Enterprise Content Management takes it to an even higher level by interacting directly at the web-services layer.
So, will iSCSI happen -- sure. Will Fiber Channel go away – not likely. Will the adoption ratio between iSCSI and FC matter much in the big picture -- not really. I believe that the value will ultimately come from moving to a SOII construct within an overall set of SOA capabilities. Sure, I bet SCSI will still be embedded in there somewhere but maybe not – maybe it will be ATA. What I do know is that it is not something that is going to keep me up at night.
For folks like EMC, we will just sell all of major interface options - take your pick. Just like selecting colors for a car – it is hard to say there is one right choice here. One year it will be green cars, and the next it will be blue cars (I think black never goes out of style). The revolution to watch, however, will be how the interaction between applications and information services changes, driven especially by the rapid adoption of SOA-based architectures. This will fundamentally change the role of and value from “storage” as we know it.