Welcome storage gamekeeper, to the guide on how to tame your monster VMs this Halloween. With more and more workloads becoming virtualised, it’s important that storage gamekeepers, or IT admins, learn to manage their monster VMs hidden deep in their virtualised environment.
Let’s start with the fundamentals - what’s a Monster VM? Well, it’s any virtual machine (VM) residing in a data center. You might call them virtualised databases, virtual desktops or something else entirely. This doesn’t mean that all VMs are evil or malicious - far from it. They have all the right intentions and when in their preferred habitats they are remarkably reasonable creatures. But when provoked or poorly placed, they all have the potential to be badly behaved. The key to being a successful storage gamekeeper, is knowing exactly what monsters you are dealing with and more importantly managing their every individual need.
If you want to identify ‘witch’ storage suits your workload and requirements, you don’t need a cauldron or a crystal ball - there is storage available that gives you visibility of your individual monsters. The question facing many gamekeepers: where should you house your monster VMs? In an All-Flash or Hybrid-Flash environment? Should you help them grow with vSphere, Hyper-V, RHEV or OpenStack? Let’s take a look at how to best tame these beasts.
Take the Gulper – this is a mission critical application, for example an Oracle database or a SQL server database. They need their own pool of I/O and VM-aware storage can do just that to ensure they (a) get the performance resources they need, and (b) don’t disrupt other VMs.
The right Hybrid-Flash solution can help to offer this however, if your Gulpers come in large sizes and numbers, then All-Flash will provide the space and I/O needed for them to feast upon. It’s important to provide a separate I/O pool for these applications so you can guarantee performance, something that legacy storage is not equipped to deal with.
If you’ve got a VM that gets a little hairy on a predictable schedule, whether it’s the end of the quarter or an especially busy retail day, you’ve got a Werewolf. Though usually mild-mannered, the Werewolf will “wolf out” at specific times. If your storage isn’t prepared, it’s hard to shuffle him around or tune him out—and you don’t want to be the one that moves a mad Wolf, do you? Using conventional storage the only option is to wait out the storm – not a great option if it’s already a full moon.
With a storage platform that guarantees VM-level QoS, you know when your Werewolf is going to snap, so set a minimum QoS threshold for that specific monster VM. This will guarantee the I/O it needs, when it needs it most. Hybrid-Flash will give you the best bang for your buck when it comes to one or two Werewolves. But if you have a whole howlin’ wolf pack, you might want to check out an All-Flash Array.
Your adorable Beast wants to be a good VM: fetching information, digging for insights, doing good VM work. But it’s a finicky fella—often changing its mood or appearance—and if you’re not pampering it, it can lose its mind. You might see these monsters in your test and development team or private cloud. You never know where or when your Beast will snap and it’s near impossible to see what set it off – leaving you troubleshooting and running diagnostics. That’s where VM-aware storage can show you the root cause of any problem across host, network and storage so you can solve it and stay out of the doghouse.
Overall you need the correct equipment to manage your monsters and in this case it’s VM-aware storage coupled with either an All-Flash or Hybrid-Flash solution. As a gamekeeper you need to create the right habitat, simplify care and feeding and make sure you keep an eye out for any spooks in the night.
Mark Young, director of systems engineering at Tintri