Rob Lopez on networks for today’s data centre
An excerpt from Transform to Better Perform, a global knowledge share initiative led by the BPI Network and sponsored by Dimension Data.
Rob Lopez, Global Group Executive for Networking at Dimension Data, outlined four key areas his team explores when planning a network for the next-generation data centre. The BPI Network shares his perspective with minor editing.
The biggest challenge facing companies right now is that they’ve got to move along with two major trends: cloud computing and the benefits they get from that. In a more mobile world, the networks they’ve already built were not designed to run and support those kinds of applications.
From a networking perspective, we stress four areas that are critical to building a hybrid system so it can operate properly across the wide area and into a cloud environment. These would complement topics stemming from the data centre itself and other layers.
The first is an understanding of switch architectures – how you connect servers and storage into a data centre, modern ways of doing it using software defined networks (SDN) and driving the level of automation. We’ve come from that environment. We’ve always built networks in the data center component. We understand the security implications, how you have got to set up virtual LANs when you set up virtual servers and storage – those sorts of things. That on-premises data centre component is critical, in terms of the way that we build and architect those services.
The second area centres on understanding what actually happens over the wide area. What are the things we can do at a network level that will actually accelerate the end user experience from an application performance point of view? That’s critical because you’re able to use networking techniques that are very application-specific to do things like load-balance or for wide area optimisation.
Increasingly, we’ve seen clients move towards delivering cloud services. We address that through what’s known as network function virtualization. We deliver those services through the cloud to give our clients on-premise and cloud applications. For example, security is something that we can spin up as a service.
Adding New Providers
The third one would be to ask how you actually manage a wider area environment to get the best possible cost and throughput and latency benefit. What we’ve seen there is that a number of our clients tended to have signed network agreements with single providers, particularly with a large brand telco that has clearing agreements with a whole bunch of other telcos.
What’s happening is that the level of telecom services provided is moving from a single provider to multiple providers. Very often, you’ll get better throughput and latency depending upon your application – just using things like broadband or VPN services – than you would out of a traditional multiprotocal label switching (MPLS) network.
If you are going to end up in an environment with multiple providers, we can take over the management of those carriers. We make can make sure you buy the right bandwidth from the right provider. We constantly arbitrage between them. This is the part of the industry that will be up for the most disruption. What we do as an independent systems integrator is to interconnect and manage contracts between multiple wide area providers.
Visit Transform to Better Perform to read this full blog post and access other insights to turn your data centre into a response-able business asset.