The impending threat of FCoE hasn’t really receded, but the good news is that it isn’t going to arrive nearly as fast as previous. It is the end of September and the FCoE and CEE standards are not showing any signs of finishing.
It is becoming clear that the impending forced integration of Storage Engineers to work with Network Engineers is going to force a culture clash of large proportions. Generally, the technology gap from FC to Ethernet is going to be substantial for many because they have come from a Server background.
That is, a substantial portion of Storage teams were Server people who re-skilled or specialised in Storage in the early phase of the FC markets development. FC is a complex and difficult protocol to master and these people have real knowledge and skills, but this knowledge is not transferable to Ethernet switching. They will need to learn new skills. A transition can be difficult.
Consider the current team integration
The Network and the Server team have a demarcation that is based on IP. The common point of “can you ping it” it usually a clear demarcation ((okay, so performance, QoS, bandwidth aren’t considered in this simplification, but run with the idea)).
The Server and Storage team have a demarcation that is based on FC. Same idea. So a Venn diagram looks like this:
This reflects that the Network and Storage teams don’t have much need to communicate.
Add FCoE – looks messy
The first deployments of FCoE will create independent networks that only carry FCoE traffic. The sales pitch of converged ethernet will get a lot of lip service, but storage people are quite definite about not sharing the data and storage networks for “lack of reliability” etc.
Which is to be expected, but it won’t take long for IT management to force the convergence issue to save money, both capital and operational. At this point, FCoE makes the team interaction look like this:
So I can look forward to meetings where there are endless debates, different points of view, and vastly different requirements. Change control is going to be a nightmare.
The iSCSI point of view.
If you take the view that iSCSI can provide Storage over IP (which it certainly can), then team interactions are going to look more like the following:
The Pain Point and its probable outcome
I can see the FCoE option is going to create significant personnel friction in the future, the combination of Ethernet and Fibrechannel will cause demarcation disputes that will take time, experience and management resources to work through. The benefits of Fibrechannel will need to be enough to outweigh this or management will stop investing in the technology and it will wither or stagnate.
The question is: Will that happen ?
Cisco and Brocade are betting on FCoE and the rest of the industry is following along like puppies. FCoE supporters point to vendor announcements and their preliminary trials. No one is pointing to IP Storage features that are getting less focus but are equally strategic. After all, old and mature technology does not make a good story.
I’m not looking forward to the clash of cultures with the Storage team. It going feel like a repeat of the mainframe networking, or Token Ring vs Ethernet, from years gone by. I predict a lot of money wasted in team meetings arguing about pointless topics, and political manoeuvring, and management involvement that isn’t going to do much to improve anyone’s life.
It seems inevitable though.