Surveillance FlashBack: Root-Cause Data Analysis


Flashback AnalysisMitigate Future Outbreaks by Determining the Root Cause of Past Problems

A key benefit of a repository is in isolating the cause of incidents that occurred behind the scenes, when the DBA is not available to experience it first hand. Many times a user calls the help desk to complain about the system performing sluggishly; but, when the DBA gets a chance to look at the system, the problem has already resolved itself. So, what next?


In the past, the only option was to wait until the problem reoccurred and hope that there was enough time to find the cause before it disappeared again. Through the use of a repository, this dilemma becomes easily resolved. The FlashBack Analysis function used in conjunction with the Historical Repository, allows IT personnel to review a period of time in which the database problem occurred as if it were in real-time to quickly analyze and determine its root cause. This is often referred to as simulated real-time of the environment. As rules-based alarming and alerting determines that there is a problem, "flashback" helps analyze and isolate the cause the problem.

 

Metrics Window

 

As the Surveillance Server Agents request collections from open Metric Windows, returned data is displayed at a refresh interval formatted as histograms, charts, etc. Metric Windows' data can be user formatted to sort by columns, select columns to display, filter the rows displayed, and groups of Metric Windows can be arranged into workspaces which can be saved and recalled.

Flashback


Metric windows are tied to one or more collections. If these collections are configured as a STORE, then the Metric Window can be viewed for any time in the retention period of the STOREs data. Much of the useful information in the Metric Windows can be reviewed for the recent past. All open Metric Windows are synchronized to the FLASHBACK date and time providing information from multiple interrelated metrics. The FLASHBACK control moves in 1 or 10 minute increments or by a specified date and time.