This is a guest post from vCloud Service Provider Logicworks which originally appeared on the VMware vCloud Blog. You can read more from Logicworks on their blog, Gathering Clouds.
Everyone is familiar with virtualization. It’s become the IT standard for achieving greater levels of resource efficiency and functionality. While it’s just a tool, the vast majority of new builds utilize it in some way.
This holds true for managed service providers (MSPs) as well. The benefits of virtualization to an MSP are similar to what an enterprise would experience. Given the nature of their business, MSPs put a great emphasis on truly being agile to client requirements, both in terms of build times and modifications of client environments.
Virtualization is absolutely key to that, and has been since the inception of VMware. The ability to resize a component of a client’s infrastructure on-demand and on the fly is an absolute must nowadays.
However, when we talk about virtualization, we mean virtualization of the compute layer, which is what everyone speaks about relative to virtual machines (VMs). And while VMs are an amazing innovation, the really interesting stuff is happening at the storage and network virtualization layers.
Logicworks is very keen on network virtualization, as opposed to the traditional configuration of hardware switches, which is a major reason we joined the NSX Beta Program.
Historically, the challenge with network virtualization’s centered on the limitations of spanning the network virtualization layer from a client’s existing virtual environment to other environments, including other data centers.
One of the major benefits we’re looking to achieve with the NSX beta program is that the technology after the acquisition of Nicira makes it possible to span virtualization between data centers which helps realize the dream of true, still completely active mobile workloads.
One of the challenges that this resolves is lead times in deployments. As it is today, providers still need to log into various switches between different vendors to configure and test them. While this is somewhat automatable, it hasn’t achieved that same degree of automation, which compute virtualization enjoys. Network virtualization gives us the ability, using software and scripts and predetermined runbooks, to deploy clients via API calls to a control cluster instead of logging into physical devices.
In addition, providers also use various vendors’ networks offerings. This means that the set of commands one will have to run on a Juniper device is going to different than on an Extreme device, and complex configurations can be quite a bit different between the two.
If we abstract that away by making the basic configuration of either of those hardware devices as simple as possible, enough to enable network virtualization on top of it, then we can standardize our configurations across our clients. This process becomes more repeatable and much quicker to deploy –like the DevOps model applied to network virtualization, to a degree. If the work being done is as close to possible from one client to another, then we can remove potential errors and increase efficiencies through more automation.
Being on the cutting edge of not-yet-industry-standard technology enables Logicworks to deploy cross-production workloads, and serve as an agile service provider. This dovetails nicely with the next generation of network virtualization in that it mirrors our ability to respond quickly and dynamically to make adjustments in deployments. For the first time, the capabilities of the technology match exactly what it is that we, as a hosting provider, do every day.