Software Defined Networking has brought a revolution to the network layer but are you able to leverage all of its intended benefits?
Nearly 15 years ago VMware introduced the hypervisor to the mainstream (I know IBM was the first) which has helped to break existing barriers in the world of compute and storage. This introduced the need for compute and storage level control (although not necessarily the solution) by separating the OS from the hardware.
Over the next decade we as IT administrators have continued to reduce boundaries and turn compute and storage into a commodity resource within datacenters. While the hypervisor revolutionized the datacenter the one thing that has historically remained untouched is the network layer.
Today, we are beginning to see, the same revolutions we have seen in hypervisors over the past decade in networking (some would argue it’s déjà vu) through software defined networking and the decoupling of physical and logical boundaries in Layer 2 and Layer 3 networking, thus introducing the need for network control.
Application Performance Issues?
Even before we begin talking about SDN, let’s take a look into how the network is managed today and more particularly how application slowness due to network latency is managed today. It typically starts with a call from an end user to an operations team about an application or service that is running “slow.” After, the “slow” application has been identified the finger pointing begins. It is then up to the network admin to identify and remediate any network bottlenecks within the environment. While the storage and infrastructure teams attempt to do the same.
Now let’s add SDN. Where before we had very well defined silos and well established paths to trace network traffic and interdependencies, Software Defined Networking eliminates these boundaries and enables machines to live anywhere within the environment.
This brings massive amounts of flexibility from an application provisioning perspective, but introduces an entire new set of complexities around an already very complex problem. The challenge with this situation is, how do we quickly and easily identify and eliminate application performance issues due to network latency? Or, even better yet, how do we optimally configure the environment to prevent network latency in the first place? This need to continuously find the most desired state of our infrastructure is what creates the need for software-defined network control.
In the typical networking environment, flow collectors and the ability to export network data via netflow and sflow is readily available. In fact, these features are even built into the distributed vSwitch in the newest versions of ESXi, all the data we need in order to effectively implement network control is readily available to us as a result.
Are your ready for SDN?
VMTurbo is able to take in and interpret all of this network data in order to identify application relationships and network topologies. Once VMTurbo builds out these relationships it is able to act as a control platform to manage the tradeoffs between network latency, demand for compute resources, and storage access.
The flexibility provided by SDN allows VMTurbo to intelligently place workloads with an understanding of the underlying network conditions in order to prevent application level latency and eliminates the need for extensive and complicated troubleshooting processes. The result is better application performance in conjunction with reduced time spent managing the environment.
The Network Control Module from VMTurbo combines management of storage and compute resources with the network in order to best manage the tradeoffs between application network latency and compute resources. In order to truly deliver a control platform for the software defined datacenter, which is needed more than ever.