As a vendor of a log management solution, we come across prospects with a variety of requirements — consistent with a variety of needs and views of approaching problems.
Recently, one prospect was very insistent on “real-time” processing. This is perfectly reasonable but as with anything, when taken to an extreme, can be meaningless. In this instance, the “typical” use case (indeed the defining one) for the log management implementation was “a virus is making its way across the enterprise; I don’t have time to search or refine or indeed any user (slow) action; I need instant notification and ability to sort data on a variety of indexes instantly”.
As vendors we are conditioned to think “the customer is always right” but I wonder if the requirement is reasonable or even possible. Given specifics of a scenario, I am sure many vendors can meet the requirement — but in general? Not knowing which OS, which attack pattern, how logs are generated/transmitted?
I was reminded again by this blog by Bejtlich in which he explains that “If you only rely on your security products to produce alerts of any type, or blocks of any type, you will consistently be “protected” from only the most basic threats.”
While real-time processing of logs is a perfectly reasonable requirement, retrospective security analysis is the only way to get a clue as to attack patterns and therefore a defense.