I can still recall, about five years ago, an industry pundit saying that iSCSI was essentially going to wipe out Fibre Channel (FC) within three years. I said to myself: '"This guy has not been in the real world."' Nothing in our
Well, Cisco and IBM were behind iSCSI, the argument went. And then Cisco surprised everyone by bringing out a FC-only director, once again demonstrating that Cisco is successful because they are pragmatic. They are not married to TCP/IP. I thought it was a brilliant move. Mind you, I think Cisco is dead serious about playing on the iSCSI side as well. But they want to make sure they can make money in it before they jump in head first.
You all remember the original iSCSI vision. Since it was based on TCP/IP (and everyone, including my mom is familiar with TCP/IP, of course), it would be simple to install and manage. Since it would not require an expensive (read: FC) switch or an expensive HBA (read: FC), it would be significantly less expensive than a FC SAN. So what if it did not perform at the same level as FC? It would be good enough for most customers and for most applications. FC can keep the lunatic fringe. All else will belong to iSCSI.
So, has this vision come to pass?
I happen to agree with a majority of the iSCSI premise. It is cheaper. I even buy the argument that more IT folks are familiar with TCP/IP than FC. Not necessarily storage administrators, but certainly the total TCP/IP pool is larger. What I do not agree with is that somehow -- miraculously -- iSCSI systems are easier to install and manage than FC systems. "You've got to be kidding me," you say. Well, here goes.
It is what the new players have done that impresses me. I will use Intransa as an example, but the logic applies equally well for others. When I looked at these offerings closely, I discovered that much of what they have done has nothing to do with iSCSI. Because these players came to the market as "iSCSI" players, it appears as if their advantages all accrue from iSCSI. Nothing could be farther from the truth. Let's take a look.
The Intransa system comes with two key parts: the system controller and the disk enclosure. Both are connected to the Ethernet switch. The system controller runs the "secret sauce" software. As soon as the units are connected, the entire disk farm is virtualized…automatically. Within minutes, I am ready for provisioning. I do need to install an iSCSI initiator on the application server. But this only takes a few minutes and is generally hassle free. It is also no cost. Provisioning takes five clicks. Yes, I suspect even my mom could manage that.
If I want to expand a volume, I can do it online and non-disruptively. If I run out of space on the disk farm, I simply add another disk enclosure, which is immediately and automatically recognized by the system controller and virtualized into the storage pool. You can allow for additional storage to be provisioned to an application without human intervention. The list goes on and on. This type of quality makes their system easy to install and manage. It is designed with "set it and forget it" philosophy. It is designed to make the job of a storage administrator easier. Connecting is easy, adding storage is easy, adding performance is easy (add a system controller) and replicating is easy. And so on. But how does one draw the conclusion that these qualities somehow automatically come from iSCSI? I would argue that they are 100% outside of iSCSI. I would also argue that these exact qualities have been requested by you from your FC players for the last seven years. You have pleaded with them to make the system easier to use. But until now, they had little incentive to do so.
OK, back to the point. While it appears that these new iSCSI players have somehow used the inherent benefits of iSCSI to make the SAN simple, nothing is further from the truth. This is the myth of iSCSI. These new players have done all the hard work that FC players should have done a long time back. All the cool features in these new iSCSI systems are 100% applicable to FC SANs. Now the FC players have woken up.
They are all scrambling to put these features in, as they also try to reduce cost. Who says competition is not good? My hat's off to these new players. In our industry, it is always the little guys who show the big guys a way to the future. I love it.
About the author: Arun Taneja is the founder and consulting analyst for the Taneja Group. Taneja writes columns and answers questions about data management and related topics.