With high-speed Ethernet, high-density fabric technology and software defined networking (SDN) all hitting the enterprise at once, it can be difficult to chart out an effective infrastructure strategy to cover the next few years. But as Brocade’s Vice President of Data Center, Switching and Routing, Jason Nolet, explains, there are ways to bolster network functionality now to future-proof systems in order to avoid major upgrades when today’s drawing-board technologies become commercial products. In nearly every case, there is a way to incorporate legacy architectures into a cloud, software-based networking strategy.
Cole: It seems that many enterprises are eager to implement private clouds, but they are intimidated by the sheer complexity of their own infrastructure. What are some of the key steps organizations need to take to prep legacy systems for the cloud?
Nolet: Enterprise customers often treat their first foray into a private cloud implementation as a mostly standalone infrastructure--a greenfield within an existing data center, if you will--and deploying a new application or service is often used as the motivator for building the first cloud architecture. VDI is one such example. I think many customers are in this ‘standalone private cloud’ state right now, with little effort being made to retrofit existing or legacy infrastructure.
Cole: As a networking company, Brocade keeps connectivity front and center. Will 10GbE really be enough for the cloud in light of the increasing use of video communications, rich media content and other high-volume workloads?
Nolet: Bandwidth demands will continue to grow as applications fill the pipe, and that’s why Brocade has been a leader in delivering 40GbE and 100GbE. That said, speeds and feeds are not really where the action is. The real challenge customers face in their data center networks is the need to simplify, automate and drive greater efficiency. Simplification and automation come from the common mandate to reduce capex and opex and to make the network as agile as the server world has become with virtualization. Similarly, the network today is inefficient, often running protocols like Spanning Tree that result in 50 percent of the links on standby waiting for a failure. That’s a very antiquated approach if you consider what has been achieved with respect to efficiency and utilization in the server environment, again with hypervisor technology.
These are the reasons Brocade has chosen to lead the industry with Ethernet fabric technology. Our VCS Fabric switches dramatically simplify connectivity by flattening the network, remove the need for most manual configuration through automation native to the fabric and double utilization and efficiency by having all links active all the time with native multipathing and load balancing at L2 and L3.
Cole: And of course, all of this seems to hinge on the successful deployment of SDN. Are there any pitfalls the enterprise should be aware of now that the technology is making its way into production environments?
Nolet: Your average enterprise customer is still in listen-and-learn mode when it comes to SDN. They really want to understand the applications and use cases for SDN and ultimately need to quantify the business benefit. The network vendors are still innovating here and it will take time for this to settle out. In the meantime, our advice to customers is that they should be choosing switching and routing products that are SDN-ready [and] capable of supporting new SDN capabilities with software upgrades rather than hardware forklifts. They should also be considering software or VM-based form factors for many of the services they run in the data center today, such as firewalls, load balancers and even L3 routing in the access layer of the network. This will save the customer money and deliver far greater agility.