Tom Wailgum wrote an interesting piece in CIO Magazine this past weekend about enterprises “still crazy for BI after all these years”. He notes that companies still see business intelligence as transformational, but he also notes, many execs are still relying on their gut to make critical business decisions. Take web site deployment for example. Even though web usage has risen to amazing levels in recent years and sites have become every organizations’ primary sales and marketing channel, we still push out new sites every day without a clue as to how they will perform in LIVE production. Don’t get me wrong, we all test our applications functionality, but rarely do we load and performance test these sites. Testing is so damn expensive. The fact is, more often then not, we don’t have a clue IF our NEW sales organization (our web sites) will actually show up to work today. Where is the business intelligence in web deployment?
The answer could be found in Cloud Testing. First, we definitely need a new, affordable, and scalable test model that moves web performance analysis away from a simple math exercise. In the past, if we had done any load testing (cost) we simulated only a fraction of the expected traffic to a site (5%-10%). Then, we would calculate what the impact of the other 95% of load would have on the site and application. A very inexact science, that has led to some pretty big and damaging headlines in the Wall Street Journal as many sites have crashed under stress and load. In another words, we need an affordable and scalable test model that allows us to regularly test 100% or 200% of our potential or expected load. Next, we need business intelligence tools specifically built to analyze the data that these tests produce. Performance testing and monitoring produce huge amounts of data that needs to be analyzed to the tiniest detail. Finding the sources of latency and stress in applications and networks can be incredibly difficult if you have the wrong tools that only provide a single dimensional view of your results such as log files or excel spreadsheets. There is simply too much test data to review. What you really need are sophisticated BI tools that provide an aggregated and correlated multi-dimensional view of the data in order to pinpoint and understand potential trouble spots. Given that our web sites are our new sales and marketing channels, web site deployment is an amazing area of uncertainty for all of us. We need a new testing model that delivers performance-based business intelligence before our sites go LIVE. We need CloudTest.
contact me at: firstname.lastname@example.org; twitter.com/lounibos