Process Availability Vs Application Availability?
As enterprises move to cloud 2.0 framework (with applications on public and private clouds), process focus will potentially replace application focus in managing such architectures. One such process-centric metric to explore is process availability.
Lets look at the way availability is calculated and monitored in the current set up
- Application Availability (%) = [1 – downtime ( non scheduled) / (total available time-scheduled downtime)]*100
This typically gets reported as below:
Here is an issue with this approach :
If a 1 hour middleware/ networking service outage (un-scheduled) happens during an end-of-period booking cycle, millions of dollars worth of transaction may end up not getting booked or reported. Availability metric however may still show a 99.99 % value.
The result : What gets reported is , perhaps, not of relevance to business.
As organizations adopt cloud 2.0 architecture, this is further exacerbated by the fact that while business processes (things that businesses care about) span applications in public and private clouds, they may be outside the purview of existing monitoring & reporting approaches.
Consider the below use case
Now lets look at how process availability as a metric may change the context of this reporting
Process : Book online order
Actual process availability for period = 99.00%.
- Out of 1 million transactions , 990,000 transactions were succesfully processed within desired SLA’s. The total value of these transactions is $X million.
- Of the 10,000 transactions that were not processed within desired SLA , 1000 were due to a middleware service outage. This outage lasted Y minutes.
- The remainder 9000 transactions were delayed due to incorrect credit information , lack of documentation or other network related issues.
This metric / report is both relevant and actionable, from a business standpoint.