Chuck Hollis, EMC VP of Global Marketing and CTO, on FCoE, Virtualization and Cloud Computing
Cisco will be present at EMC World 2009 and on the web at “Live at EMC World” Bill Marozas ,Cisco Director Storage Networking, took this opportunity to ask Chuck Hollis, Vice -President Global Marketing, and well known EMC blogger tough questions “Chuck Hollis here from EMC. I was asked to “guest blog” for Cisco on a few topics that I found interesting. I hope you find the questions (and answers!) thought provoking! Q: What is your view of the evolution of the Storage Networking industry? In particular, how do you see the use of transfer protocols evolving in the data center? I think we’re all pretty agreed on the end state of storage networking: Ethernet wins – period. It won’t happen overnight, but the outcome is inevitable, if you think about it. The more interesting questions revolve around the “how” and the “when” of this transition, rather than the final outcome. Right now, enterprise data centers are intrigued about the potential of converged Ethernet data center fabrics for their next generation architectures, and in many cases are building it into their long term plans. As a result, I think you’ll see a healthy co-existence of FC and FCoE for many years – traditional storage networking infrastructure working alongside newer converged Ethernet infrastructure using the same switches, management tools, etc. Very few customers would ever consider a rip-and-replace approach. A related debate is “what storage protocols will we see on this converged wire?”. For us at EMC, this really isn’t a religious debate, and more a case of customer preference driven by specific use cases.
For smaller environments, we’ll still see plenty of iSCSI on 10Gb Ethernet – simply because the IT people are comfortable with setting up TCP/IP networks and don’t really want –or need — to learn all about the world of FC management models. Content-rich and file-sharing environments will end preferring NAS over their Ethernet wire – there’s no arguing that managing everything as a file system is a pretty attractive model for many customers. However, the big applications in the big data centers – they’ll inevitably gravitate to FCoE. They like the predictable characteristics, they like the partitioned management model, they’ve invested a lot in understanding FC and how it works – and they don’t see any real reason to downgrade to another model, since the economics are largely the same for all three protocols when considering converged fabrics. The beauty of converged data center Ethernet is that you can do all three – and more – simultaneously over the same wire at the same time without stepping on each other’s toes, so to speak. It’s easy to imagine consolidate VMware farms where some are accessing files through NAS, and some accessing high – performance databases through FCoE, and using the same converged adapter to do inter-server RDMA for example. And that’s a pretty attractive end-state from the architect’s point of view. Q: What’s the role of storage networking in data center virtualization? There are several interesting levels to this question. As a starting point, consider than in a fully virtualized environment, all the resources need to be fully virtualized (and dynamically orchestrated!) in order to participate. This includes server, network and storage – and, of course, storage networks specifically. This means that we’ll need to be able to do things like dynamically set up and tear down storage connections as things move around, or have storage connections (and associated attributes) follow around virtual workloads as they move, or perhaps dynamically vary the bandwidth and response times up and down as conditions shift. We’re doing most of that with today’s technology – one recent example is how EMC has recently brought dynamic path management to the VMware world (PowerPath/VE) — but many of us can see an opportunity for doing things better in the not-too-distant future. I think things get much more interesting when we start talking about creating virtualization pools that span data centers and time zones. To be truly useful, information has to follow the workload as it moves around, and we see a wonderful opportunity to offer up some very innovative technology to do just that in a cost-effective manner. This sort of geographical perspective makes you start to consider topics like storage networking and storage virtualization very differently in the near future. Q: What are the top challenges facing customers considering data center virtualization? I have a controversial perspective on this – it’s not so much about technology, it’s about the “3 P’s”: people, process and politics. Even today, the technologies from VMware, Cisco and EMC are capable of doing much, much more than most people are actually using. Virtual data centers can be thought of as private clouds – they’re built differently, they’re run differently, they’re measured differently, they’re funded differently – and we have several decades of IT thinking that has to evolve in order to fully exploit what the technology is capable of today. Unfortunately, we think enterprise IT organizations are going to get caught up in a squeeze play in the next few years. Just as outsourcing created a new external benchmark that was used to drive internal IT efficiencies, we believe as more and more service providers start to stand up “virtual private clouds” to provide infrastructure as a service, platform as a service or software as a service – IT organizations will be forced to realign there organizations and their technology platforms to close the gap between what can be done internally vs. externally. Thanks for inviting me to blog for you! Chuck Hollis VP – Global Marketing CTO http://chucksblog.emc.com