Reinventing the Data Center
David Spark reporting for Riverbed at Interop '09, Las Vegas, NV.
On Wednesday, Interop in Las Vegas opened with a keynote discussion entitled "Reinventing the Data Center." The moderator was Art Wittmann, Managing Director, InformationWeek Analytics, and joining him on the panel were:
- Dr. Mark Stuart Day, Chief Scientist, Riverbed Technology
- Paul McNab, Vice President, Enterprise & Mid Market Solutions Marketing, Cisco
- Dave Stevens, Chief Technology Officer, Brocade Communications Systems, Inc.
- David Yen, Executive Vice President and General Manager, Data Center Business Group, Juniper Networks
Read their full bios at the bottom of this page.
Here's a summary of the issues and points brought up in the discussion:
- To the question, "What equipment should you first purchase?" one person answered to first think about services and creation of those services. Another replied, you need to know your communications and switches.
- The data center has become a less siloed environment. Being able to provide dynamically available computing is where we're striving for (e.g. cloud computing).
- The data center is an ecosystem problem when you're dealing with solutions from so many vendors. You would think vertically integrating would be the solution, but that locks customers in.
- There's more competition to innovate on products rather than race to create standards. But an innovation may lead to a standards body.
- Greatest needs for standards are needed around virtual machine mobility and transport. More importantly is the need to maintain quality of service once that information comes out of its "compute node."
- Fibre Channel over Ethernet (FCOE) - Don't want to get caught in a situation where only certain vendors can work with each other.
- It's going to be multi-protocol for a while. We need to hide that translation confusion.
- FCOE deployments are happening at the edge of the network. We want to drive that conversion deeper into the network.
- The FCOE growth will happen before you actually see storage traffic going over it. The data center will want to implement it as a reserve to allow for growth.
- It takes three to six months to deploy applications because there are so many different networking and storage people you need to coordinate before you can test and launch anything. The ideal situation would be to have them siloed for their expertise, but integrated in a managerial sense so you can turn them on in a dynamic way instead of having the delay of coordination.
- FCOE is not a driving issue. People are struggling with virtualization, cloud computing, and storage duplication.
- Network architecture is not just dealing with issues of connections within the data center, but also those connections to disaster recovery, to the cloud, etc.
- Consolidation is one of the areas you can immediately see return on investment. The cost savings will mostly be in reduced power consumption.
For more, check out all of Riverbed's Interop '09 Las Vegas coverage.
Comments