Getting the Big Picture – Part One
A while back we talked about how the face of monitoring has changed. Historically we monitored individual servers and their performance because we knew so much about what servers we had, and what software was running on them. Today with the cloud or virtual servers your load can move without your knowledge or influence.
Today we monitor service performance. However, in order to get the big picture you may need to monitor multiple services instead of multiple servers to understand your complete application performance.
Your customer calls in with less than acceptable performance on your web site. The feature they are using uses services from a Web Server serving up the web pages, business logic executing on WCF, Data Access running on WCF, and persistence in an SQL Server.
All of these services may be hosted on many different configurations of hardware. Any layer could co-exist on the same server, be distributed on a separate server for that service, or be deployed on multiple servers, clusters, or virtual machines for that service layer, etc.
Now that your customer calls in with their problem, how do you track down where the performance bottleneck is? Is it a network issue? Is it a database issue? How about load balancing, or do you have one host in your service that is not performing as well as the others?
In two tiered systems we simply blamed the database first, and then moved to the application with things tanked. As your systems become more distributed, simply blaming the database is no longer effective.
I’d like to explore how you instrument your applications and configure monitoring to get the big picture allowing you to monitor your system to establish baseline performance, alert on performance degradation, and use monitor data to diagnose reduced performance.
This is a great topic for those of you with experience in this need to leave comments by replying below, or sending an Email to btaylor@sswug.org.
Cheers,
Ben
$$SWYNK$$
Featured Article(s)
Tips for Minimizing Deadlocks
Deadlock or fatal embrace is a situation, which occurs when the first process is waiting for the resource that is locked by the second process and the second process, is waiting for the resource that is locked by the first process. In this article, we will learn about ways to minimize deadlocks.
Featured White Paper(s)
Extending the Value of SCOM with Spotlight on SQL Server
read more)