The storage area network versus network-attached storage technology debate is a "religious" argument that is given too much play, says Meta Group researcher Kevin McIsaac.
People should take their eyes away from the technology and concentrate on how they use the technology, says the US-based researcher. Cost, ease of management and availability are the things that matter, he says; their ranking depends on what business you are involved in. A bank will naturally want availability above all else, whereas a small business will probably concentrate on cost. A dot-com will consider time-to-market paramount. The right solution is the one that provides the right mix of those benefits. Unfortunately, each solution attracts its zealots, particularly among technical people, he says.
At a technical level, SAN and its associated fibre-channel architecture are suited for transferring blocks of data – typical of database work – he says, while NAS (network-attached storage) and internet protocol deal efficiently in file-sharing. If most of an organisation’s work is centred on databases, then SAN is the logical answer. If it is file-centred, then NAS appears the way to go.
McIsaac was speaking at an EMC-organised roadshow touring New Zealand and Australia. While proclaimed as independent, his “horses for courses” point of view subtly supported EMC’s release of the 4700 storage interface device, which exists in both fibre-channel and IP versions.
Taking on one technology does not preclude using the other in parts of the organisation’s operations where it might be more appropriate, McIsaac says.
The business considerations come down to four basic principles, he argues, and these should be approached in order:
- Separate the storage from the servers. This allows the business to run different replacement lifecycles for storage and servers, rather than the one being tied to the other. That in turn makes for more rational decisions and lower total cost of ownership.
- Consolidate storage, so you avoid having one crammed server and one with space going begging. This also eases management, which means saving on people – always the biggest expense.
- Storage can have its own dedicated purchasing and maintenance staff who deal with specialist storage vendors – not vendors of servers with a restricted range of “round brown stuff” already attached. Tape and backup strategies should be included as an integral part of the storage planning and automated at much as possible using a centralised library - again a saving on people. Storage can then be packaged as a service, with defined service-level agreements.
- Only when considering the interface between storage and the network do we touch on the SAN v NAS debate, he says. Infrastructure change projects are costly and hard to sell to management, McIsaac says. It is better to wait for a definite business-inspired applications project, and ask “why don’t we try SAN/NAS with this one?” Then once it’s in one part of the organisation, it can be gradually spread throughout the networks as other new projects emerge.