Reducing storage pollution

If we believe the predictions from various sources, hybrid cars will graduate this year from being toys for eccentric people to mainstream vehicles for gas-efficiency-conscious drivers. This cheery forecast is based in part on major car manufacturers, including Honda, stating they are readying new hybrid vehicles including SUVs and sedans.

This is great news. Hybrid vehicles lower gas consumption, which would seemingly help reduce that dreadful plume of polluted, brownish air suspended over our cities’ skylines, an all too common image.

But the widespread acceptance of hybrid vehicles, like any new technology, faces an uphill battle. It will probably take at least a decade before these vehicles are popular enough to have a significant impact on the air quality in our cities. Even so, they will likely never completely replace gas-fueled vehicles, which should maintain their validity for more demanding uses than the simple work commute or driving to the mall.

This discussion about hybrid vehicles brings me to another instance of competing technologies: iSCSI, the transport protocol for networked storage based on TCP/IP, and FC (Fibre Channel), the incumbent.

As is the case with hybrid cars, iSCSI is just now trying to prove its worth over an already well-established technology. Nevertheless, this promising networking protocol seems to have a better chance than hybrid vehicles at quickly gaining market share over the incumbent. 

In fact, according to a recent study published by IDC, in 2004 iSCSI vendors worldwide will sell a respectable $1 billion worth of disk arrays, five times more than in 2003. That figure is expected to rise to a staggering $3.2 billion in 2006.

I would rather not comment on those numbers, but I certainly agree with the general idea that one can draw from those estimates — there is a large market segment waiting to buy iSCSI products.

Interestingly that research, based on users’ surveys, projects a significant growth of iSCSI-based arrays for both networked and DAS (direct attached storage) deployment. 

Considering that initial FC implementations were mostly for DAS, which requires much simpler and less worrisome scenarios, I am not surprised. However, deployments of iSCSI DAS promise to be even easier and less destructive.

For instance, users can add iSCSI storage to a server over a built-in GbE (Gigabit Ethernet) NIC, without installing a new adapter card in the machine. In addition, an administrator can use existing GbE cabling to move storage arrays away from the server, for example, to a more protected, vaultlike room.

Deploying networked storage requires more user knowledge and awareness than deploying DAS. But the intuitive, familiar nature of iSCSI is often less intimidating for many IT-savvy technicians than FC can be. This means the iSCSI protocol may very well be easier to learn and manage than FC, but it will still require some education.

That said, I believe training and technical evangelism will probably be the major challenges facing iSCSI vendors in 2004. Not surprisingly, the Storage Networking Industry Association (SNIA) has seen fit to start the year with on-the-road proselytism in the form of seminars about IP storage held in four major cities.

Check these seminarsout and, if at all possible, try to attend. You could find out that IP storage can solve real problems at a reasonable price, and that it can be more environmentally friendly. But bear in mind that it will not and should not completely replace FC in the same way hybrid cars will not replace traditional gas-guzzling vehicles anytime soon.