A View from the Rainy Northwest Kier Gombart's Weblog

Sun Connection Availability

I pulled together some data on the availability of our applications over Q3 to make sure our customers were getting the service they need, and to measure the effectiveness of our Operations and Engineering teams in delivering top notch service availability for Sun Connection applications. 

We set a target of 99.5% availability as the lower spec limit (and strive to blow this out of the water every month).  The data in these charts is different than standard url pings, as they are run from 5 to 9 global Internet sites, using a service from Mercury.  The measurements represent a user logging into the application, or accessing a service (commonly referred to as customer experience monitoring).  So, ALL unavailability, including Internet congestion, service outages, or even planned maintenance, is reflected in these numbers.

Overall, we are doing pretty good.  For most services... we are four 9's or above.  The Sun Update service seems to just stay around 99.7+%.  We've had issues, mostly Tahiti eligibility availability, some database failures and problems with JMS messaging, that impacted this number.  Over time, we've been adding monitors (that is why some are marked n/a). 

The performance graph below shows that the services have been responding consistently over the month.  The load on the production servers has been low (mainly due to the faster servers we migrated to last year), so I would expect these to stay in the normal range.  The top line seems high, but is in actuality a cumulative measurement of 4-5 screens (called a deep transaction).  We do this test to make sure the internals of the app are working.  The lines on the very bottom, are most likely just pings to verify the service is alive and responding.  Over time, we'd like to make these more complex to better measure the performance of that service.

For more info, please see our quarterly metrics site.

thanks...Kier