Cracking the iSCSI conundrum

What’s going on here? How can a major storage architecture that gives servers access to big chunks of disk space have so many pitfalls, that it achieves “dirty secret” status at precisely the time the industry is starting to seriously propose it to customers? The answer is quite simple, it’s because iSCSI uses conventional networking components.

Cracking the iSCSI conundrum

All those holes you see on the backside of the PS4000 are regular Gigabit Ethernet connectors, and when you set up your server to talk to the PS4000 – or, indeed, to any other iSCSI responder device – you do so by feeding in an IP address via a neat little control panel applet in Windows Server.

To get the PS4000 going, you plug into its first available interface and run through a web-based setup procedure, and even at this high end of the marketplace there are still lots of signs that this is just standard networking business.

The final proof for me was that once we got down to final configuration and assembly, all those leads you can see trailing around the edges of my terrible photo, connecting the iSCSI device to the server, came straight out of my “network onsite bag” – a disreputable grab bag of CAT-5 cables accumulated during years of misadventures.

This would appear to be the basic source of iSCSI problems. Not the cables themselves, but interoperability. Take a look around the lower end of the iSCSI marketplace and you’ll find that a number of vendors are happy to leave all discussion of this particular stumbling block to techies in their support forums; you know the kind, where the minute the right answer crops up that thread shrivels away to nothing.

Dell switches

Interoperability is the key to making the whole technology accessible for us, and it’s also the best way for any vendor to deliver something that people can try out on the learning slopes, without spending tons of cash on a leap in the dark.

That’s how we ended up with USB drives, to drag this discussion right down to the bottom end. USB is a widely understood standard, and the use of USB storage has taken off because you can just plug them into the same hole you use for your mouse – that’s what I mean by interoperability.

And it’s all well and good while you can still reach those files on the USB drive, until you come across one of those mice I’ve seen that sends your BIOS off to some very dark place and stops the entire machine from booting.

Now iSCSI has its very own version of this kind of problem, but one that’s more subtle and even more insidious. When setting up this PS4000, a Dell switch and some Dell servers to converse with one another, I made use of nothing but regular network technologies (and I was ably assisted by an excellent chap from Dell who was clearly well used to carrying monstrous bits of enterprise hardware about in shopping bags on the Tube).

We plugged the storage array up to the servers using wet string from my spare CAT-5E cable stash, and we set up the servers’ network cards by ambling about within Windows 2008 Server’s somewhat over-decorated network card configuration dialogs. If you’d just sat down at those machines after we were finished you wouldn’t have been able to easily tell that this network was set up for iSCSI – that is, for exposing an array of drives to a machine equipped with an initiator – that’s how neat and tidy the use of standard network tricks is.

In fact, the only reliable guide to the disastrous nature of interoperability in this field is the sharp intake of breath it receives from industry experts.

Disclaimer: Some pages on this site may include an affiliate link. This does not effect our editorial in any way.

Todays Highlights
How to See Google Search History
how to download photos from google photos