Datacenter Automation

Leading Web search portals are using semantic search engines to deliver answers instead of results. That same technology is now emerging in the enterprise. It can help developers tame log data by uncovering information about application performance problems and answering the question ‘what went wrong?'.

We've all been there: you're looking for something on the Web, and the search engine returns a lengthy list of blue links. All of the Web pages are shown in one place, but it takes time and effort to find the right one. The same is true for many application log tools, which may centralize information but fail to identify the actual problem.

Enterprise applications can "live" in many places and their logs might be scattered and unstandardized. First generation log analysis tools made some of the log data searchable, but the onus was on the developer to know what to look for. That process could take many hours, potentially leading to unacceptable downtime for critical applications. Proprietary log formats also confuse and confound conventional keyword search.

That's why semantic search can be so helpful. It uses machine intelligence to understand the context of words, so it becomes possible for a Google user to type "cheap flights to Tel Aviv on February 10th" rather than just "cheap flights" and receive a listing of actual flights rather than links to airline discounters. Bing Facebook, Google and some vertical search engines include semantic technology to better understand natural language. It saves time and creates a better experience.

The time saving is even more dramatic for developers who are using log analysis tools with this capability. The challenge of reading proprietary logs is immediately solved by virtue of the semantic ontology, so it's no longer necessary to examine each console separately. Machine intelligence also augments the ability of the developer to find answers within minutes instead of hours. It's called intelligence amplification.

U.S. military researchers first conceived of intelligence amplification during the '50s and '60s. It's now being put into practice by applying semantic search principles to apps.

Application log searches that are augmented with intelligence amplification can prioritize events, and zero in on the proverbial needle in the haystack. It becomes possible for a developer to quickly learn that a database connection, rather than an internal application issue, was the root cause of a failure. A keyword search may turn up hundreds of connection problems when only 15 are really meaningful.

After having been proven on the Web, semantic search is emerging as a valuable technology within the datacenter to save time and solve problems for developers.