Feature

Converging technologies

Much has been made of the rivalry between Nas and San. In reality though, Chris Youett, they are two complementary technologies.

There has been a sharp outbreak of peaceful co-existence in the storage market this year. Following the threats of a nuclear exchange between the Nas and San camps over the past two years (much to the amusement of the direct attached storage device, Dasd, suppliers) the warring factions have been pulled back from the brink by their business partners and independent consultants.

Quite simply this was because most sites were much more interested in a strategy which would consolidate their disparate storage devices. Prior to Y2K, most DPMs would have been hard-pushed to say just how many disk drives their organisations had.

Now they know - which was often two or three times more than they thought they had - DPMs have been told by their boards to rationalise all this storage and impose proper business discipline, such as quantifiable returns on investment (RoIs).

Many DPMs had been guilty of taking their eyes off the ball over the past couple of years, especially in the storage area. This had led to a proliferation of technologies and devices. These can be summarised under three headings: Dasd, Nas and San.

While Dasd is still the most popular, especially for the mass of traditional IDMS, IMS, ISAM and VSAM data, San is now seen as the strategic way forward, with Nas slotting in nicely for those application areas which need 'plug 'n' play' storage.

Mike Harper, product marketing manager for Storage Tek EMEA, says: 'I have never seen Nas and San as competitive architectures. We can support Nas via a San infrastructure. We find that San currently tends to be good for campus environments.

'Nas has the advantage of supporting IP, which makes it better for longer distances. I expect the two technologies to start to merge when we get Fibre Channel across IP. This should happen during the next two to five years,' he adds.

Rumours of the death of Nas have been put around by San enthusiasts for some time. Mark Kinsell, Nas specialist at Hammer, believes that Nas will have a place as long as it maintains significant cost of deployment savings and ease of integration over San.

He adds: 'What sites need to understand is that each technology appeals to different markets. In the short term Nas will gain market share from traditional Dasd faster than San.

'Nas is currently good for those sites which recognise that they have network blockages. If you can pinpoint these, and add a Nas box, you can reduce local contention, and improve network throughput. I expect this approach to be very popular, as it takes over the next six months. However, it won't start to make real inroads into Dasd until manufacturers can deliver very high reliability,' Kinsell points out.

Nas also offers freedom from the traditional production server environment, notes Doug Rainbolt, Exabyte's director of sales and marketing. He adds: 'By adding Nas boxes you can free up your production servers from file serving, thus improving performance overall.

'Administrators favour Nas because, unlike dedicated backup, they can better centralise and manage the backup and restore activities. However, there are limits to deployment, especially if you try to transfer boatloads of data across a sacred production network.

'In the realm of 24 x 7 operations, either this needs to be seen as a transitionary step, or one should plan on putting a cap on potential growth.

'You can get around this to some extent by using a thin server Nas device. Nas can present a way forward for those sites who want to rationalise all their disparate servers only in the short term. This is because the pending bottleneck of excessive data traffic will sometimes raise its ugly head,' Rainbolt says.

Nas will also have a future while San is perceived to be expensive and requiring a minimum of £100K of infrastructure, notes Keith Balding, technical director at Primary Storage.

He adds: 'Nas is currently very simple to install and administer. It adds I/O processing power, but is not good for resellers, as the hardware is almost a loss leader.

'However, the technology is excellent for mixed sites as it only has a single partition. It can also help to reduce backbone problems, especially if the site is moving big Tif or JPG files around,' Balding says.

The other major trend that sites can expect over the next two years are boxes which can support both technologies. Nigel Houghton, Northern Europe sales manager at Gadzoox, adds: 'I expect Nas and San to merge into network appliances. They are complementary technologies. For example, Nas is good for sharing files while San is block based and takes you forward from SCSI.

'As NAS boxes can then access those blocks, don't be too surprised if the rest of the industry follows MTI's and LSI's lead of offering boxes which do both - especially when sites start to find that scaling up their NT building blocks becomes very costly.

'Nas is ideal as the next step, especially as many popular software packages are based on the likes of Oracle and SQL Server, which also uses block structures internally,' he says.

Mark Hiley, sales and marketing director at LSI's logic storage systems division, notes: 'Each type of disk storage has its place in IT. While Nas isn't really suitable for high I/O applications it is great for sharing data and files between different systems - and I expect there to be a big take-up by Internet Service Providers (ISPs) and e-biz.

'As the underlying technology is often the same high availability disks, I expect to see Nas and San converging over the next three or four years. We can offer this now,' he adds.

Fellow dual box leader MTI is also frank about the sorts of applications which Nas isn't suitable for. European business development manager Roger Bearman adds: 'There are draw-backs in the management and control of Nas. This is why you probably wouldn't put TP applications on to it.

'Nas technology is ideal for starting e-business - and you can move to San when trading volumes grow. We offer one storage operating system to manage everything under the covers, and we use our own patented technology with Seagate 73 Gbyte disks,' he points out.

As Dasd is still the market leader, when did Bearman expect Nas to take off? He replies: 'We have spent the last 18 months shipping Fibre Channel technology - but there are still a lot of hearts and minds to be won over from traditional Dasd.'

Wider adoption of Fibre Channel technology is seen by many suppliers as the key to Nas success. So what does the trade association think? Paul Talbut is chairman of Fibre Channel Industry Association Europe. He welcomes Nas providing it is used appropriately.

'The aim of Fibre Channel is to overcome the limitations of SCSI, so that sites can have mainframe-quality bandwidth with the flexibility of Lans.

'We map SCSI, IP, and Escon onto a common level. This gives consistent frame production, and allows enhanced speed and distance between servers and storage. This delivers more fault tolerance, and makes it easy to plug in Nas boxes.

'At our European conference (June 5-7 at Maastricht) we expect to see 2 Gbit per second over 10 Kilometres Fibre Channel announced. By 2003, the standard could be up to 10 Gbits per second over 30 Km.

'This will allow most sites to use Nas and Ethernet on the front end with San on the back end, without affecting user bandwidth to any server,' Talbut predicts.

The other development which should also set the Nas world alight is being closely monitored by LSI's Hiley. Most pundits expect SCSI over IP in the next three to five years.

Hiley concludes: 'SCSI over IP? Watch this space!' l

The Microsoft option
Open up most Nas or San boxes and you will probably find the same Raid (redundant array of inexpensive drives) inside. Microtest believes that instead of wasting time with sterile arguments over which technology is best, it has opted for an open Raid system called FileZerver, which can be configured for most common storage tasks.

Jim Selby, Nas product manager at Microtest, points out: 'Most other suppliers offer closed boxes, and assume everything you need is inside. We offer a controller which you can build around what you want.

'This gives sites, especially in small to medium enterprises, the scaleability they really want from their storage sub-systems. We can support up to 19 hard drives giving between 200 Gbytes and 1.4 Tbytes per controller.

'In the real world, groups of users have different requirements. For example, you can have high volume data creators, such as clients or networks running e-commerce, Cad/Cam, imaging, or audio/visual applications with high capacity and low cost secure storage.

'There are also workgroups within a larger network who can share common files and want scaleability on demand, independent groups who want peer to peer networks, and educational groups who want data that is easy to access, but is centrally backed up and kept highly secure.

'The biggest threat to Nas is the trend to slip in embedded NT for administration. This is a 'no way back' option, especially for administration,' he adds.

So how does Microtest achieve this flexibility? Selby replies: 'With the humble paperclip! There is a hole in the baseplate, and you stick the end of a paperclip in to reset to the manufacturers' specs. We support most SCSI, Mac, Unix, and NT configurations natively.'

SUMMARY
The industry no longer sees network attached storage (Nas) and storage attached networks (San) as rivals, but as complementary to each other. Most suppliers and software houses expect the two technologies to merge over the next five years with a 'horses for courses' approach being recommended to sites.


Email Alerts

Register now to receive ComputerWeekly.com IT-related news, guides and more, delivered to your inbox.
By submitting you agree to receive email from TechTarget and its partners. If you reside outside of the United States, you consent to having your personal data transferred to and processed in the United States. Privacy

This was first published in May 2001

 

COMMENTS powered by Disqus  //  Commenting policy