Tuesday, March 27, 2012

Do you have ‘Actionable Intelligence’ (AI) on your Virtual Infrastructure?

I have been doing a lot of vCenter Operations and have to admit I really like the version 5 product.  A little less enamored about the licensing but only because of where they but the demarcation point between Advanced and Enterprise.  Let me share a few thoughts with you and explain the AI title.

Actionable Intelligence is something I picked up from interacting with VMware’s sales folks.  It is a key differentiator between vCenter Operations (vOPS) and many other systems management tools. 

The secret sauce is in the dynamic thresholds.  Rather than setting a static threshold such as CPU over 75% utilization (as you get in vCenter), vCenter Operations configures dynamic thresholds.  Every 24 hours, algorithms are run to identify what is normal behavior and what is abnormal behavior on the objects that are aggregated in vOPS.  The  objects and data Is pulled from vCenter. 

Why is this so unique?  Well the proof is in the patent.  All the algorithms are patented from VMware because they have incorporate their intellectual property on what’s best for your vms.  VMware, in the development of their systems management products noted that in many tools the data is difficult to interpret or potentially suspect.  In the design of vCenter Operations, considerable time was spent to ensure the information and the display is straightforward and provides information you can act on (hence the ‘actionable intelligence’ buzz word).

Does it deliver?  Well you need to allow time for the technology to recognize normal and abnormal behavior.  For example; if vOPS is running for 1 day and an irregular spike in CPU on a vm occurs it may register an anomaly.  If this spike happens every week at the same time then it will be considered normal behavior.  VMware generally recommends that you run the tool for a break in  period of two weeks.  The real-time monitoring will be useful the moment it is installed however.

One of the nice integration points is the inclusion of the Capacity IQ reporting and what-if modeling in a ‘single’ product.  I quote single as it is still two vms (the analysis and UI vappliance) released as a ‘single’ vApp. 

My only minor point is on the licensing break; essentially standard is real-time monitoring, advanced is everything including long term risk and efficiency.  You are restricted in vOPS advanced to a single locked dashboard which is the default console.  Enterprise allows you to unlock the customizations so you can build separate dashboards.  Enterprise ++ allows you to incorporate physical and virtual machines. 

What I find is that the minute it is deployed, customizations are critical.  You may want the management team to only see a high level view vs. being able to drill down on all critical alerts for example.  To move from Advanced to Enterprise is a license key upgrade on the same product which unlocks the customization utility.  It’s almost like they planned it this way to entice the Enterprise upgrade, hmmmmmm…