Standards in storage networking.

'Standards must be good, everyone has one', or so the saying goes. The unfortunate reality is that this is the case; however there is some light at the end of the tunnel--at least in one area. A few years ago the Storage Network Industry Association (SNIA) started down the long and sometimes tortuous path to creating a standard for storage management. Initially codenamed 'Bluefin' the draft standard was created by a number of vendors who had recognised the problems, and had come together to do something about it. A survey by 'The Info Pro' in September 2003 found that for 54% of customers, improved management tools was still their #1 storage issue. When the draft was complete, it was handed over to the SNIA, who promptly adopted it and opened it up to a much wider community for review. Thus the Storage Management Initiative (SMI) was born.

By creating a common management specificaton (the 'S' in SMI-S), vendors were signing up to enable common management. The first version of the specification was released and while it was primitive in some ways, it was a huge step forwards. The most common tasks had been standardised and companies from IBM to EMC to VERITAS have put aside their competitive nature to reach agreement on how these tasks will be done.

However, just having a standard defined is not good enough, the next steps are to get adoption of the standard and roll out products which support it. This has been happening over the past few months and now there are more than 100 products from more than a dozen vendors on the market which support the standard. The next version is in progress and the launch of a certification suite (Conformance Test Program, CTP) has help drive adoption.

So why was it done? Vendors are notoriously uncooperative when they are competitors, but it essentially it comes down to two things. It is good for the customer and it is good for the vendor, also it is because of this that SMI-S will win out. Customers will be able to look for the SMI-S compliant label on their hardware and software purchases and know that it will fit into their environment seamlessly, without the need for new management applications and the training that goes along with them. From the vendor side, instead of having to negotiate with every other vendor that they wish to interoperate with, they can write to a standard set of APIS. This will enable them to bring new products to market faster and reduce the costly development of proprietary APIS. The next version of SMI-S expands the scope of management to cover more functionality and covers more of the hardware that makes up a storage network, e.g. improved HBA manageability as well as resolving things like multiple paths to storage correctly. Standards die unless they evolve, we are seeing SMI-S evolve and now that products are reaching the market, we will see stronger adoption by even the smaller players, who just can't afford to remain proprietary any longer.

While some standards are innovating in the way which common management can be achieved, others are pushing technology to improve network speeds. Fibre channel standards are created and pushed under the auspices of T 112. They have standards on the drawing board that are looking to take us through to 10Gb in 2008. While 2008 seems a long way off, in computer terms this is actually quite close. The technology needs to progress to a point where it is possible and then the standards need to be applied. In some ways, from the outside, standards seem to move glacially slowly, when inside it can often be a roller coaster ride. Other protocols such as ISCSI have the same tough milestones, all of which will benefit the consumer in the long run. By pushing the technology it pushes the standard, which in turn makes vendors create products with improved performance and features but which don't mean the wholesale rip and replace of existing technology. Will standards ever lead to a truly open architecture or truly open systems? I believe the answer is that it will from consumer pressure. Consumers made the move to open systems because they didn't like the tie-in with proprietary systems and in the end they voted with their feet, or rather with their chequebook. Storage is now at a point where standards are emerging and words like 'open' are being applied to it, those vendors who do not comply with standards will fall by the wayside. However, standards do not mean the end of innovation. Vendors will still have the opportunity to show how good they are and how much better their product is--but by using standard mechanisms. SMI-S is based on CIM (Common Information Model) which can be extended by vendors to meet their needs. The extensions will, over time, become standardised and part of SMI-S, but all the while vendors will continue to innovate and extend the model. The commercial attractiveness of a proprietary solution, with a high price attached to it has gone. Standards have won.

Guy Bunker, Veritas

www.veritas.com