Service provider takeaway: Storage SOA could bring a host of benefits to resellers and their customers, but a few pieces to this puzzle are still missing.
Here's the problem storage SOA could fix: The storage world is drowning in features such as replication, compliance, continuous data protection, backup and archive, and each supplier has a unique way of handling them. The major storage manufacturers provide many of the capabilities, but there is no single platform that can do it all and, more importantly, do it all well. Adding to the din, there are many startup companies that exist simply to address one of those features.
Storage SOA promises to quiet the cacophony by establishing a common platform for pluggable storage resources. Storage SOA would not enable these features; rather, it would provide a better platform to host the services and a better interface into those new features.
Here's how storage SOA would work. Your customer, Buy and Large Corp., buys storage products along with a series of core services from its storage vendor. Sometime after this storage is installed, Buy and Large decides that it wants a feature such as versioning. To enable versioning for Buy and Large, you would simply install it on the company's existing storage system.
By enabling services to be easily pluggable, storage SOA would produce unimagined solutions and capabilities. The level of innovation would be unprecedented because it would provide software companies with open access to the various storage platforms, removing the risk of having to reinvent the storage foundation itself. This would also give software companies more time to develop new features instead of re-creating old ones.
If such a scenario were to come to fruition, storage SOA would be a boon to resellers, small software companies and customers alike. Resellers would always have a chance to provide additional services to a customer. With today's model, once a customer has made a storage purchase decision, the possibility of providing additional products and services to that customer drops significantly, especially if you're not the first provider with a foot in the door. In a storage SOA model, selling plug-in enhancements to a company's storage system would be much easier, since the system could be sold incrementally rather than in a whole-hog fashion.
From the perspective of small software developers, storage SOA would bring a great advantage. For example, they could develop a better snapshot engine than the original storage supplier, or they could develop features such as continuous data protection (CDP), compliance or block-level information lifecycle management (ILM) -- all without having to install an additional appliance that carries integration and support challenges.
For customers, storage SOA would mean that there's no need to buy an appliance for capabilities beyond those in the core storage array -- and, maybe better yet, no need to wait for the supplier to pony up the features itself.
The problem with the nirvana that I'm describing here is that this model may never happen on a scale large enough to make a difference. Storage SOA would require broad implementation at the foundation, which today is owned by the primary storage suppliers. The technology is there, but the desire on the part of these market leaders is not.
First, I doubt seriously that any of the major suppliers are going to open up their architecture to accept these modules from outside sources (also known as competition). The current market leaders in the storage hardware space fancy themselves "software manufacturers," and they are not likely to allow other software suppliers to integrate into their platform.
All hope is not lost, however: By resisting storage SOA, storage hardware suppliers are leaving the door open for switch manufacturers to enable it. In fact, this is already happening. Brocade and Cisco both have some ability to host applications on their intelligent switches. CDP, storage virtualization and replication solutions have all found their way to either Brocade's or Cisco's intelligent switch frameworks. While these switch solutions are not true storage SOA -- too much proprietary work is required to create the modules -- they are the closest example of storage SOA that we have.
For storage SOA on switches to be successful, intelligent switches would need to have a more rapid adoption rate and a more open development architecture. This would mean a lower price or a higher value (which might mean a wealth of services to choose from). And, the storage software manufacturers would need to be comfortable running their applications on these switches. Brocade and Cisco would compete less on speeds and feeds, which in many data centers are becoming irrelevant, and more on their software. To attract more software suppliers to the storage SOA game, switch integration would need to be easier.
Is storage SOA dead on arrival? Not entirely; the concept will make good business sense to your customers, but like with ILM a few years ago, the storage SOA market will take a while to develop. For now, it is merely an interesting subject. There are simply too many pieces of the puzzle that need to fall into place before storage SOA has any hope of being successful. There may be components of the architecture to explore now, such as offerings that plug in to your favorite switch provider, but for the most part my recommendation is that storage resellers sit this one out a while.
About the author
George Crump, founder of Storage Switzerland, is an independent storage analyst with more than 25 years of experience.
This was first published in February 2008