It's the Elasticity, Stupid

By Lisa Crewe, Sr. Product Marketing Manager and Ran Gilboa, Technical Director, NetApp


Have you ever thought about whether to buy or lease a car?  The decision whittles down to your own situation based on a variety of variables, but you have to know the variables.


The cloud is  similar to leasing and it’s expensive.  It’s also expensive to move back and forth from on premise to cloud, so you should have the intelligence to make the best decisions.   In general, an elastic cloud application or process has three elasticity dimensions, cost, quality, and resources, enabling it to increase and decrease its cost, quality, or available resources,  to accommodate specific requirements.


So, how do you get this information?  What should you move?  Will it cost more or less?  And once the  move is complete, how do you measure the elasticity?   Most of us will start with the basic variables – applications or VMs, capacity, cpu, IOPs, cost, etc. this is a good start but it doesn’t tell us anyting about how elasticity of the IT infrastructure, or the potential of adopting elasticity. And how do you get visibility into all of these variables to weigh the pros and cons?


The following is an example of a Computing Elasticity Dashboard that you could create with OnCommand Insight storage resource management software to monitor these variables and help you make those decisions.  Let’s review some of the content in the dashboard to show how it can help you.




First, let’s look at the VM’s Allocated vs Deallocated.  You can see by the graph below that there’s a spike in activity in the November  time frame. For the purpose of this example, let’s say this is a retail organization and the spike can be attributed to Black Friday.  Otherwise,  the activity is similar throughout the year.  We can conclude that this may be a good candidate to burst into the cloud during that timeframe.

Now let’s look at the lifespan of the VMs over different time periods.  We can see that a large majority of the VMs are only used for less than a month.



The next section describes the savings potential.  More than 60% of the VMs lifespan is less than 3 months, this indicates elasticity adoption.  What’s left is to identify not the VMs, but the projects that are elastic.

Finally, we can use the information to determine which projects  are good candidates to move to the cloud.  First, we need to understand the different projects.  You can see they are listed in the graphic below.   As an alternative, we can cherry pick the specific VMs, but this will likely be tedious work. 


Why don’t I don’t see any elasticity in my IT infrastructure?

In most cases,  you haven’t implemented “reclaim” methodoligies. Here are two ways to identify opportunities to “reclaim” by looking at the VMs.

  1. The first way is Inactive vs Idle, vs Operational.  If VMs are inactive or idle, there’s a good chance they can be reclaimed. 
  • Inactive – suspended or shutdown
  • Idle – less then 5 CPU and 5 IOP/s
  • Active – all others

2.  The second way is by  looking at all of the VMs and looking at whether they are CPU bound or Disk bound.  If they are neither CPU or Disk bound, they are a likely candidate for reclamation.



As we’ve shown, having an executive dashboard to monitor elasticity variables is key to getting the most out of your cloud implementation.  Without it, you won’t have the necessary intelligence to make informed decisions to assure that you’re getting the most out of your hybrid cloud strategy.


How do you measure elasticity in your IT infrastructure?  We'd love to hear your comments.