This article can also be found in the Premium Editorial Download "Storage magazine: Managing data storage for remote employees."
Download it now to read this article plus other related content.
I don't think so.
What's becoming clearer is that new technologies will have to meet a demanding test: Can they solve a pressing problem now? Style points won't be awarded. Nowhere is that clearer than in the contrast between two technologies that are currently hitting the market, InfiniBand (IB) and IP storage.
Cisco's unveiling of its IP/Fibre Channel (FC) switches means that iSCSI finally has a big-league champion in the market. But that would be only mildly interesting if it weren't for the fact that our surveys consistently say you're more interested in it than any other emerging storage technology. This coming year may well be the Year of IP Storage, with iSCSI playing a big role.
Sure, the first wave of adoption isn't likely to be in intense applications like those used on the trading floor. Initially, at least, the payoff will be to network functions like disk-based backup, off-site disaster recovery and non-mission-critical storage. That's all good.
This is a technology that was built for doing more with less. In this issue, you can read about how Carlson Companies, with many years of high-end
FC isn't going away next year, nor should it. But iSCSI has the potential to change the SAN cost equation in a big way, creating a tier of lower-cost storage networks. Its IP heritage may also lead to a new respect for interoperability. That way, you could mix and match iSCSI cards, routers and switches on the same network, something FC doesn't always do well.
Whether iSCSI can deliver in production, day in and day out, remains to be seen. Also to be determined - mostly by you - is whether or not IP storage traffic will get its own pipes or share the LAN, or some combination of the two.
In stark contrast to IP storage, InfiniBand has not gotten a rousing reception from storage managers, and consequently major vendors have cooled to it. You'll be using IB, no doubt, to connect servers in clusters. And some of the arrays and blade servers you buy will have IB inside the box. But the vision of a single IB network linking all data center devices - worthy as that is - seems on hold until the day when IT budgets have a line item for "elegant solutions."
Innovators in these lean times still need to have one hand on the paradigm shifter, but they also need to think in terms of moving the speedometer through tweaks and incremental cleverness. In the late 90s, it was often fun to think in grand terms. Now, progress demands the art of the practical, even when it comes to innovation.
This was first published in September 2002