Back to Blog

Yuri Rabover

IT Agility is Vital to Survival

In IT as well as games survival of the fittest depends on IT agilityThe fittest survives

Why do people prefer to deal with the same entities over and over? They're predictable, they look the same, behave the same, you don’t need to think as much. Imagine that all apples and oranges are the same size, you can order the same boxes and you will always know how much it will weigh. They all have the same proportions of vitamins and sugar and acid, everything you would need.  Nice and predictable. Heaven.

But what if you need to feed children during the flu season with fruit with more Vitamin C so they could be more resistant? You will probably need to buy more of these standard apples or oranges and spend more money and time on packing them. Consider this in nature. Only the species that can adapt to changing conditions survive – and conditions change constantly.

The world of IT is similar. One size doesn’t fit all – frankly it doesn’t fit anything as you don’t know what this size should be. There are many types, classes, sizes of pretty much anything you have to deal with. If you look inside any modern data center you will find at least 4-5 hardware vendors, dozens of software vendors and hundreds of various types and combinations of stuff. It is never identical, and if you think of it this is actually good – you can always find the right combination of products, tools, configuration to fit your goal to deliver predictable service at optimal cost. So agility is extremely important in order to survive in today’s IT. Let’s look at some aspects of IT agility in more detail.

Why is there more than one hypervisor in the datacenter? After all, if one is technically superior why would anybody use anything else? For example, consider ESX and Hyper-V. They offer pretty much the same functionality but ESX has been around for much longer and considered more enterprise ready. Moreover, they are both free, so this should be no brainer, right?

But let’s look further. Certain functions you absolutely need like vMotions or storage vMotions you cannot perform on ESX, you need to buy a license of vSphere – and not just a simple standard license but more advanced. Without vSphere, you cannot group hosts into clusters, you cannot use resource pools, templates, etc.  You also need to buy something extra for Hyper-V but Microsoft structured their tools entirely differently – features are packed into different product sets and have different costs. And if 90% of your load is Microsoft anyway, you could be better off getting all tools and support from a single vendor. But what if performance is worse than with vSphere?

So as you see, even with such a simple example the choice is not obvious at all. It depends on the total cost of ownership and your ability to deliver services at a predictable cost. It could be the case that you may need more than one hypervisor to optimize these parameters and be able to dynamically move the load across them adapting to the changing conditions: load, cost, demand.

Let’s continue this example. You realize that the cost of maintaining a large vSphere environment no longer fits your business needs, the margins become too thin. You start planning to switch to Hyper-V. It is OK to start with a little pilot, 3-4 Hyper-V hosts, no other management software. You started a dozen VMs, they all work great and the next question is how much of Hyper-V you need to move 30% of your load off vSphere.

First, you need to know the size of the target hosts and datastores to fit these 30% of the load. Second, which particular 30% of the load would be the best candidates to move to Hyper-V? Third, are you going to replicate the same cluster structure in vSphere with Hyper-V, or do it differently? And once you figured it all out you should be able to continue running it on a day-to-day basis, this is not a one-time activity. Next month you are thinking of moving extra 20% of your load but then some of the load could be moved back from Hyper-V to vSphere. Which one?

These are just very obvious and straightforward questions with regard to only one aspect, managing workload across 2 hypervisors. The answers may not be so obvious. IT agility is important, but are you equipped to handle it? Do you need a solution?

Image Source: