In the first part of this article I looked at how, with a well-architected product, it is perfectly possible to run both NAS and SAN (file and block) storage from a single storage pool. Why the confusion then, and why are storage companies still selling separate infrastructures?
One of the main reasons is that many vendors simply do not have the ability to offer converged NAS and SAN, and therefore feel the need to hide their products weaknesses in marketing waffle. Vendors have developed two separate product lines to address file and block data storage. In most cases, converging them would mean throwing existing architectures out and starting again. According to Mike Walters, consulting systems engineer at Net Apps:
Many storage vendors have no capability to present NAS and SAN from the same controller, even if the SAN is iSCSI (Ethernet-based). Hence, this might come across as a you cannot do this feeling.
Whether this is passively coming across, or is being deliberately portrayed as a you cannot do this feeling, probably depends on how cynically you view the situation.
Storage vendors stuck in the old world also have to tackle the issue of: why sell one box if you can sell two? Naturally they are reticent to invest heavily in a new product architecture that cannibalises 50 percent of their own or their partners business.
In addition, there is the question of selling disk. Many storage vendors, subsidise the cost of their systems by using technology to lock customers into purchasing overpriced proprietary disk for capacity upgrades. If customers have two separate storage subsystems for NAS and SAN, vendors sell more disk, as Kirill Malkin, CTO of Reldata, clarifies:
At any point in time a storage system will have a certain amount of redundancy: disk which has been bought and paid for, but which is not in use. With two separate storage subsystems for block and file storage, there is likely to be twice as much redundancy. For a vendor whose margin comes from selling proprietary disk this is obviously very helpful for cashflow. If these companies enable customers to rationalise capacity across NAS and SAN storage they eat into one of their major sources of profit.
Whether it is due to inept technology, lazy product design or a desire to maintain stronger disk sales, by unnecessarily maintaining the need for two subsystems, vendors are creating additional costs for the customer. And these additional costs are not only in terms of superfluous product and unnecessary disk purchase. According to Roger Turner, director of storage systems EMEA at Hitachi Data Systems:
One of the major benefits of a converged storage subsystem for block and file storage is in consolidating administration workload and minimising management cost. In many cases this may deliver greater savings than from hardware rationalisation.
By storing all data in one place, not only management but back-up and continuity planning is massively simplified. Therefore, vendors who push separate systems for NAS and SAN, are not only over-selling equipment and disk, they are unnecessarily driving up their customers administration costs into the bargain.
Even outside of the storage subsystem there is still further unnecessary separation between NAS and SAN. An FC SAN is clearly a different beast to any form of IP storage. Currently, adding NAS or iSCSI storage functionality to any FC SAN requires additional storage servers, even if these are simply blades mounted in a rack. But the industry has used this inarguable difference to create an exaggerated separation between NAS and newer iSCSI-based IP SAN technology.
From the customers point of view, there is no good reason for separate products to deliver IP SAN and IP NAS. A good IP storage product should have no difficulty in handling iSCSI, CIFS and NFS protocols from one device, comments Reldatas Malkin. In an enterprise, capturing the benefits of IP storage will require storage servers or gateways to be distributed across the organisation in branch offices and separate departments. There is absolutely no need to buy and maintain two devices at each location, anymore than there is a need to maintain separate networks for voice and data. That is the whole point of converging on IP.
So why has the storage industry been waffling about NAS/SAN convergence for so long, with so little to show for it?
The answer is that convergence of any kind is rarely a win:win scenario. A large number of suppliers usually end up losing out. Customers will gain but at the expense of many storage vendors. Currently, however, vendors still clinging to the status quo, maintaining unnecessary divisions between NAS and SAN, are costing their customers far more than the ill-gotten revenues from the sale of unnecessary equipment and disk.
Rory MacDonald, Technology Creative
T: +44 (0)23 9281 4259
M: +44 (0)7899 965 232