If data center management is your thing, you know what a headache it can be to try to keep up with everything that is really happening on your servers. IT monitoring software is prevalent. Which applications are running where? How much traffic are they really generating? Can you seriously lockdown security and still keep everyone up and running with the tools they need every day? And can you really see what is going on without pouring over tons of data?
It has to be a manageable operation. With the amount of data that passes through the servers every day, it would be an overwhelming task to sift through all of that data to find the information that your need, especially when it comes to possible intrusions or other security issues.
The Usual Flow Data is Great, But…
Machine learning is now being applied to the data center environment in order to improve the visibility into the realities of the server environment. Instead of using the customary packet sampling techniques, IT monitoring software that uses machine learning sifts through ALL packets. Normally, this would place a remarkable burden on the servers; however, the only information that is collected is the header, leaving the rest of the packet untouched. The rest of the packet remains intact, also serving to protect encrypted data as well.
One of the unique aspects of this type of collection and analysis is that it is done in two places – the header is captured on the network, and the analysis is performed at the server level. The server will know which application dependencies to map. Does it create big data? Of course. But it is the machine-learning in the system that brings the two scenarios together to allow the necessary visibility into the data center, matching the headers to the dependency maps and returns packet travel, protocols, applications, etc.
What Else Can I Do With Machine Learning?
Machine learning is promising to be useful for migration planning, using dependency mapping to find servers that are no longer being used (and can therefore be re-purposed). One of the more interesting advantages of IT monitoring software that incorporates machine learning is to determine any adverse affects on applications once any changes to the network security policies are made. Those are a few of the capabilities that machine learning is proving; the coming year may bring physical infrastructure discovery, and/or further security features. To learn more about machine learning, check out this free lecture from Stanford:
If you enjoyed this blog post, please check out our other blog posts on IT Monitoring Software