With everything in the IT world being virtualized to optimize cost and induce savings, multi-tenant data centers are the reality of today. These multi-tenant data center run thousands of applications which vie for the same infrastructure and in-turn the infrastructure is guided by configurations (by operators), making it a complex environment which stands on the three pillars; Application, Operator and Infrastructure.
With such complexities the slightest of changes in the work load or configuration can have devastating (cascading) effects and result in performance issues or revenue loss, as was learnt by Amazon in the April 2011 outage of their AWS service which resulted in an estimated revenue loss of approx 2 million.
Another study on data center pegs the per minute outage cost at $5,600 to the providers. It goes without saying that appropriate diagnosis of problems and behavior is of essence to ensure availability and keep check on the cost of running a DC but just a reactive or even a predictive approach is not enough to ensure smooth operations and therefore companies need to move to the next level, i.e., prescriptive. It is essential to analyze the relation between the three pillars of the DC because it is this interaction which has the most unpredictable effects.
To predict the unpredictable is the name of the game for Big Data. Solution providers need to utilize the big data platform and marry their assurance and operational solutions with big data analytics to predict the multi layer interaction in the data centers. A lot of platform providers like Cisco, VMWare and Huawei are looking to move into this space with a unified solution for data center operations, while at the same time a few new providers (like Perspica Networks) with niche skills have forayed into this domain. The advantage of these niche providers is that they are not coming with a baggage of moving from well established legacy platforms into the predictive or prescriptive world, but their products have been designed from bottom up to work in coherence with the analytics platform to provide a seamless prescriptive solution to the real world problems of maintaining and running a data center.