Tech Field Day – #VFD4 – VMTurbo Putting a price on resources!

VMTurboLogoSmVMTurbo closed off the first day at VFD4 in Austin, Texas with an overview and deep dive into their flagship product Operations Manager.  This was one of the presentations that I was most looking forward to as my fellow Toronto VMUG Co-Leader, fellow Canadian and good friend Eric Wright was involved in it, and for the first time I got to see Eric on the “other side of the table” speaking for a vendor.

 Disclaimer: As a Virtualization Field Day 4 delegate all of my flight, travel, accommodations, eats, and drinks are paid for.  However I do not receive any compensation nor am I required to write anything in regards to the event or the sponsors.  This is done at my own discretion.

Demand-Driven Control for the Software-Defined Universe

Eric started off by prompting everyone’s thoughts around what exactly Operations Manager is– not by talking about the product or what it can do, but by briefly explaining a motto that VMTurbo has been built around – Demand Driven Control for the Software Defined Universe.  I know, it’s a long one but in essence it’s something that is lacking within the industry.  With the Software Defined X being introduced into our data centers, it has brought with it many benefits, and perhaps the biggest being control – we can now have software controlling our storage, software controlling our network, and in the case of automation, we have software controlling our software.  And as Eric pointed out this control is great, but useless if there is no real consistency or leverage behind whatever is controlling it – and in fact, having the demand, having our infrastructure be the driving factor behind this control is truly the answer.  VMTurbo’s Operations Manager is a product that helps us along our path to the Demand-Driven Control for the Software-Defined Universe and it does so in it’s own unique way…

Desired State – Datacenter Nirvana

herethereBefore we get into VMTurbo’s unique take on operations management I first want to talk a little bit about desired state.  Looking after a virtual datacenter we are always looking to bring our VMs, our applications, our workloads into what we consider a desired state.  This desired state essentially combines both availability and performance together, all while maximizing the efficiency of the resources and infrastructure that we have to work with.  We begin to see anomalies and performance issues when we veer away from this desired state, and traditionally, we, as administrators are tasked with bring our workloads back into that desired state.  VMTurbo states that this is where the problem lies – this human interaction both takes time – time for humans to find out about this shift, as well as time for humans to try and put the puzzle back together and get back to the desired state.  VMTurbo essentially takes the human interaction out of this equation – allowing software, in this case Operations Manager to both detect the shift from desired state but also, and more importantly take action towards moving your environment back to the desired state – thus the “control” part of the Demand-Driven Control.

And the Demand-Driven part?

This is where we see the uniqueness of VMTurbo’s Operations Manager shine through.  With Operations Manager in control, making the decisions of what VMs should run where, etc. it needs a way to look holistically at your environment.  How it does this is by taking an economic model and applying that to your infrastructure, essentially turning your datacenter into a supply chain.  Every entity of your environment either supplies or demands resources, and just as in economics when there are a lot of resources available, things are a bit cheaper.  As resources go down, things begin to get a lot more expensive.

VMTurbo-Marketplace

So in terms of a VM demanding resources Operations Manager calculates the cost of those resources, again, holistically across your entire environment to determine just how those resources should be provisioned.  Think of adding more disk to a VM –  you need to look at where the disk will come from, how expanding that disk will affect other consumers (VMs) on the same datastore, how the extra capacity will affect other suppliers such as your storage array, your LUN, etc.  Operations Manager calculates all of this information in real time to determine how to best provision that storage capacity to the VM and takes action if need be to free up resources or create more supply, all while maintaining the desired state of all of your applications.

Operations Manager also goes deeper than just the VM when determining who it’s buyers are.  Through the use of WMI, SNMP, or by simply importing metrics from a third-party tools Operations Manager is able to discover services inside of your operating systems and throw them into the crazy economic market as well.  Think of things like Tomcat servers, Java Heaps, SQL Server, etc.  These are processes that may affect the demand for memory, and without insight into them making a recommendation for more memory on a VM isn’t going to help anything.  By taking these granular metrics and statistics from inside of your VMs operating system, Operations Manager can give a complete recommendation or action that will best suit your application, VM, and entire infrastructure.

It still does all the other stuff

Now VMTurbo’s supply chain model definitely sets it apart from other monitoring tools.  Also, the fact that Operations Manager can take action automatically also is a big plus when comparing the product to other tools but  you may be asking yourself what about all of the other stuff that most monitoring tools do today?  Well, Operations Manager does that as well.  Items such as right-sizing a VM, taking away or granting CPU to a VM, placement, capacity planning, etc.  Operations Manager does all of this an in fact it also applies these actions to its supply chain model, allowing the software to see just how granting another 2 vCPUs to a VM will “disrupt” the market and decide whether or not that change is “worth it”.  Operations Manager also has some decent networking functionality built-in as well.  By figuring out which VMs are “chatty” or communicating each other often, Operations Manager can make the recommendation to move these VMs onto the same host, eliminating any performance degradation or latency that could occur by having the communication move out across your network.

When VMTurbo takes action it does so in a manner of a recommendation or an action – meaning we can have the software recommend the changes to the user or we can have the software go ahead and take care of the issues itself.  Honestly this is a personal preference and I can see customers probably using a mix of both.  When calculating these recommendations and actions Operations Manager also places a transaction cost on any move it makes.  What this does is alleviate VMs from essentially bouncing back and forth between hosts trying to achieve their desired state.

Operations Manager really looks like a slick product which takes a different stance on monitoring and healing your infrastructure.  Having the application that is doing the watching do the actual doing makes sense to me, eliminating the need for the human interaction which in turn eliminates risk and certainly increases the time it takes to get back to desired state.  And I know I’ve specifically geared this post towards vSphere but honestly VMTurbo supports just about everything – think OpenStack, Azure, Hyper-V, AWS, vCloud – it’s got them all covered.  If your interest has at all peaked I encourage you to watch all of the VMTurbo #VFD4 videos here – or better yet, get yourself a trial version and try it out yourself.  Oh, and this just in – get your name in on a home-lab giveaway they are having in respect to their newest launch.