Why Texas Tech University switched Fibre for iSCSI

I have been writing about iSCSI for several years now, frequently taking the position that iSCSI technology is complementary to (and not, as had earlier been assumed, a competitor of) Fibre Channel storage-area networks.  I have long believed that the two SAN technologies would compete when Fibre Channel becomes cheaper and simpler to use, leading them to compete for IT dollars within the same market niches.

Common wisdom (“common” does not, of course, equal correct) has held that Fibre Channel will always own the high-end SAN space at enterprises (with their well-trained IT staffs), and that iSCSI is best applied to smaller and less demanding storage requirements where the staffing may not be as sophisticated. Why then would a company that already has a large and well-trained IT shop that’s already running Fibre-based SANs install iSCSI devices? 

Here is how the decision-making went at Texas Tech University, which recently made just that choice.

At the outset, the university’s departmental Web and file servers had only direct-attached storage.  As a result, adding storage required server downtime, with the administrator physically configuring each disk drive on each server.  In addition, Texas Tech had no business continuity strategy to get it through either planned or unplanned outages, so every time the servers came down for storage upgrades the users were out of luck. 

What the university was looking for was a solution that physically consolidated its storage, scaled easily and minimized back-up and restore times.  Texas Tech also wanted its new storage to be inexpensive, reliable, flexible when it was time to allocate to space, and be manageable from a centralized location.

The decision to go with an iSCSI SAN incorporated each of the following considerations:

* The university liked the ability to consolidate storage for many of the school’s Web servers and departmental applications on a SAN.

* The university liked the fact that a SAN is a scaleable solution, allowing it to add more storage (in this case, SATA drives) as needed and seamlessly.

* The customer determined that with a SAN it would be able to reduce operating costs through eliminating the need to purchase new servers, by simplifying storage management tasks, by reducing the number of administrators it required, and by reducing downtime.

The device the university chose – an iSCSI Storage Server from FalconStor – used Microsoft’s Windows Storage Server 2003, which provides a familiar Web user interface to the management console. This meant less learning time for administrators and thus faster time to value.  It also provided the users with data replication and recovery capability between the primary site and a disaster recovery site.

Remember that this site was already running Fibre Channel.

Clearly, the site knew the value of SAN-based storage, and understood their need to consolidate storage and centralize its management.  It appears as if the decision in favor of the iSCSI SAN rather than Fibre Channel was driven largely by the iSCSI device’s familiar management console and by its disaster recovery capability.

This example illustrates two important points.  First, it shows willingness on the part of IT managers at larger installations to consider iSCSI-based SANs and Fibre Channel-based SANs as competing technologies.  Secondly, it shows that the decision-making had a whole lot to do with how easily manageable the solution was and that the kind of pipe the solution ran through was not much of a consideration at all.

If this continues to happen at other sites, it will mean that technology decisions regarding “the pipe” have been relegated to a much lower level of importance than the business decisions involving operational expenses.  That maybe a good indicator that it is time for the industry to rethink its positioning of iSCSI and Fibre Channel.