More and more often we see various parts of IT landscape being outsourced: Datacenters and WAN’s, Servers and Firewalls,- you name it. When doing so, IT managers would typically rely on provider’s monitoring tools of as a mean to detect service interruptions. What is almost always lost in such cases – is the ability to monitor services end to end.
Consider this typical situation that I came across very recently:
One morning a database guy complains about about synchronization issues between two datacenters. Both locations, as well as WAN, are managed by providers. The IT support had to log in into 3 provider’s portals to discover the following:
- Their availability reports are built using different metrics and non-matching scale;
- Their availability reports are not granular enough;
- Their availability reports do not cover everything you need. For example, they may display availability but not utilization.
Guesses? The reason for failure can be anything not even appearing on provider’s graphs: database engine glitch, overloaded server, spanning tree loop inside datacenter, routing convergence issue, QoS drop,- you name it! We will never know what actually happened.
Results? As they love to say in movies, they are “measurable and reproducible”. Comparing with proper monitoring setup:
- Time to detect the source of problems is much higher;
- Ability to analyze things properly is much lower;
- Ability to stop fighting fires and do instead predictive analysis is next to zero.
With all these nice outsourcing options, one service which I think should always be present – is the central E2E monitoring system. Either buy it as a service or keep inside, but never lose it.
Alex Mavrin, CCIE #7846
Visit http://www.apteriks.com and use FREE ONLINE tools for network professionals.