The four principals of Private Cloud – Part 2 – Service Analytics
While defining and publishing a set of IT services that meets or even exceeds the expectations of your end user to help them improve the agility of the business is the first step in delivering IT as a Service, we have to be able to make sure that we don’t over-promise and under-deliver.
In order to keep our promises, the IT governance disciplines we have developed over the years around capacity planning, and infrastructure design and deployment remain important but are transformed as we focus more on service optimisation. The luxury that we had in the past of assuring SLA conformance by over engineering and over-provisioning within an application silo is no longer affordable. More to the point this siloed mentality often leads to a lack of standardisation across the different silos which makes subsequent automation incredibly difficult. On the other hand in our drive for lower costs and greater agility, we cant sacrifice the security, reliability and failure isolation that siloed infrastructure gave us, and which IT infrastructure professionals care so deeply about. The kinds of shared virtualised infrastructures that cloud computing is built on must not only give us greater efficiency and flexibility into our environment, it must at the same time give us the ability to increase our level of management and oversight, and improve our ability to take corrective action if it is required.
The old adage that “You cant manage what you cant measure”, still holds true, but the trouble is that measurement in highly virtualised IT infrastructures presents a number of challenges. In larger IT organisations no one group holds all the information or expertise to troubleshoot performance problems, or to identify which resources are being consumed by a given user or application. We cant assume that because that, for example, the network, compute and storage resources are all meeting their SLA targets that the end user experience is also acceptable or meeting the SLA offered in the service catalog. Having team leaders sit in a “Come to Jesus” meeting with their arms folded saying things ike “It cant be the storage teams fault, we bought the most expensive frame array in existence and over-engineered the hell out of it, it must be a problem in the network” simply doesnt cut it in a world where service levels are king and nobody cares who’s fault it is, or which widget is currently fubar.
The bestway to address this is to make appropriate investments in the tools and processes that goes beyond simple monitoring. These tools need to enable IT staff to do things like identify service paths and confirm the redundancy of those paths, set policies on service paths for accessibility, performance, and availability, intelligently analyzing logs and other data to ensure and report on whether policies are are being adhered to, and and provide capacity planning forecast intelligence to allow optimal use of resources and implement and leverage just in time purchasing processes.
By building a solid Service Analytics capability, you can immediately improve overall governance and lay the foundations for successfully managing and optimising private cloud services via
- Consumption-based metering
- Dynamic capacity optimization.
- Reduced management and resource costs.
- Provenance and Audit-ability of SLA conformance