In our last post on the Infrastructure Convergence Continuum, we focused on the Build Your Own / DIY Architecture for virtualization infrastructure. There are architectural limitations with this implementation that we addressed in the first post (“the inverted pyramid of doom”) that may be worth reviewing as a baseline understanding for today’s post. Why? Spoiler Alert: They share the same architecture as the Reference Architecture and Converged Architecture we’ll be covering in today’s post.
Products such as EMC’s VSPEX and NetApp’s FlexPod are considered reference architectures which are inherently the same as a Build Your Own / DIY model we described in our first post. In fact, they are so similar that I debated whether or not it was even worth putting in an image to compare the two. See below and you’ll understand why:
The only difference in the two sides above is that the infrastructure on the right has been “pre-certified” to run as a proven architecture with specifically identified components in the stack. This gives the IT generalist the ability to quickly implement a known solution based on well documented use cases. As with the DIY Architecture, this is a tried and true method for providing high availability. The problem is that it still relies on the complexity of each layer being managed independently. It is a step toward convergence, but really falls short when it comes to reducing the complexity of implementation and ongoing maintenance and support.
This next category of solutions in the convergence continuum is the Converged grouping. This classification is typically associated with VCE (VMware, Cisco & EMC’s partnership) products, but products like Dell’s Active System would fall into this category as well. This solution has the High Availability benefits of a DIY solution, the added benefit of the pre-certified nature of the reference architecture, and combines it in a single SKU to be easily consumable from a purchase perspective to the midmarket IT generalist. The marketing bundle lessens the time to implement over a reference architecture and gives the user a perceived “one throat to choke.” In reality, the complexity of each independently managed piece in the DIY architecture still remains as does the propensity for the independent vendors being packaged together to point fingers when it comes to supporting the overall solution.
For those of you who read the original post covering the DIY architecture, this title should look familiar. Again, since Reference Architectures and Converged Solutions all share a common architecture, they all share the same fatal flaw. They rely on multiple servers and hypervisors having the ability to share a common storage system, which makes that system a critical single point of failure for the entire infrastructure.
The turning point for these architectures has been the ability to combine both storage and compute processing within the same physical hardware, which provides an opportunity to greatly simplify the deployment of the overall solution as well as reduce costs vs. discrete physical components and complex inter-networking.
Two primary architectures have been used (VSA Model and Hypervisor Convergence Model) and will be described in our next two posts. First we’ll cover the VSA Model of convergence and then we’ll compare that to our Hypervisor Convergence Model. Stay tuned!
Please complete the form below and someone from Scale Computing will be in contact with you within 24hrs.×