For months people have been asking me what I’m doing, and it’s been difficult to hold something like this under my hat because I’ve been really excited about this.
How the time has flown. I joined Cisco in June of last year, and on my first day my manager told me that I was going to be working on a new product -- an FCoE blade for the MDS 9500 Series Fibre Channel Directors.
Talk about drinking from the firehose! In this case, it was a lot like attempting to drink from several hoses at the same time. It quickly became apparent that I was working with some truly incredible people, people who not only were doing the engineering for the product (and it’s related software) and running it through its paces, but also those who were directly involved with me in determining how it fit into the Big Picture™.
We’ve talked about end-to-end FCoE before, not to mention how FCoE at the access layer is a great starting point for virtualized servers (especially when it comes to cable reduction). We’ve also talked about the different ways of extending into Multitiered and Multihop environments.
Even according to this criteria, Cisco has been able to do true Multihop with the Nexus 5000 Series of switches.
What we haven’t talked about is why Director-Class Multihop FCoE is so important.
Options, Choice, and Agility -- Oh My!
This is what has been promised since the first conversation of FCoE began waaaaaay back in 2007. This is the ability to truly converge the data center and yet, at the same time, bolt it on to existing ones. The brownfield/greenfield war is a false dichotomy, because FCoE allows you to implement convergence where you need it.
As Consolidated I/O targets become commonplace, you have a choice between connecting with the Nexus 7000 series or the MDS 9500 series Directors, allowing the core Fibre Channel capacity to connect to the FCoE access layer while preserving all the reliable network services and features you’ve come to expect.
I’ve said it before, and I’ll be saying it again, I’m sure: Cisco’s implementation of FCoE is fully standards compliant, standards that have been completed for a while now, and you don’t need any other type of standards (not TRILL or QCN, e.g.) in order to accomplish your goal of end-to-end consolidated I/O throughout the Data Center. You can do this with all the visibility you’ve come to expect in your Fibre Channel SANs, including zoning, management tools, and scale.
Putting My Money Where My Mouth Is
During the past several months I managed to run Early Field Trials (EFT) for Director-Class FCoE. Over the next coming weeks I’ll be describing more about this, but for now let’s just say it was the most comprehensive and involved EFT Cisco has ever done. Encompassing Nexus 2000, Nexus 5000, Nexus 7000 and MDS 9500 switches, Fibre Channel and Fibre Channel over Ethernet hosts and Targets, spanning both the US and Europe, we managed to touch nearly every aspect of the Data Center on a small scale using Multihop FCoE.
One of the key things that pleasantly surprised many participants was how simple it was to reconfigure traffic flows according to needs. It happened exactly like I predicted: the first time administrators were able to traffic engineer according to configuration settings rather than adjusting physical cables - and the broader implications of this ability -- they were hooked.
A Feeling of Accomplishment
I can’t begin to say how relieved I am that I can finally talk about the project I’ve been working on for all this time! I’m also honored and humbled by the fine craftsmanship of the team that has developed these products with a single goal in mind -- giving the customer the choices they need to make the most reliable, efficient, and agile data centers possible.
Words cannot express how proud I am to be part of this, to be a Product Manager for FCoE for Cisco’s Data Center Business Unit and present to you these products that will be the foundation for changing the way we think of the Data Center. The team I’m involved with have gone above and beyond the call of duty to make FCoE synchronous across platforms, interoperable, and rock-solid.
This is only the beginning.