* What's the driving force behind EMC's plan for enterprise domination?
Today, we finish our overview of where EMC is going. Last time, we were describing the SMARTS technology that EMC acquired to give it root-cause analysis capability across the enterprise. You will recall that for storage, SMARTS uses data collected by ControlCenter (ControlCenter gets its info using both SMI-S and proprietary interfaces).
ControlCenter normalizes the data and then passes it up the line to the SMARTS InCharge software. At some later point, the data will be munged together with data from other areas of the enterprise to give managers a comprehensive enterprise overview. Many companies are today using pieces of InCharge, but from the EMC presentation I saw, it was impossible to tell to what degree.
InCharge is still a work in progress, of course. Data collection depends on “adapters” (ControlCenter is the storage adapter), and in cases in which adapters have not yet been built, no information gets into the system. Thus, the speed with which these adapters can be brought online will play an important role in determining what SMARTS can be smart about. If you are looking for value in understanding the cross-domain relationships that this software potentially reports on, find out about the schedules for those adapters that will be most meaningful for your sites before you make any decisions.
Based on what I heard during the meeting at EMC, InCharge is not rules- (or, if you prefer, “policy-“) based, but rather relies on looking at the network connections and then interpreting events by looking at the symptoms that the events have spawned. At the heart of all this is a good old-fashioned table look-up that correlates problems and symptoms. InCharge first interprets those symptoms based on the likelihood of certain problems being the cause; then comes the nifty part (and, presumably, the SMARTS intellectual property), in which probabilities are weighed to select a likely root cause.
Probabilistic approaches such as this are somewhat rare and, because they rely on statistical analysis, are likely to have inherent shortcomings in cases in which the causes of a particular set of symptoms are statistical outriders (that is, they represent an unusual case). In other words, the rarer the root cause, the less likely it is that the root cause can be identified. This may not be much of a problem in established networks, but may be much more problematic in more volatile situations.
Hopefully the SMARTS correlation technology can account for this – after all, SMARTS has been working on it for quite some time. At this point the software has no heuristics for learning on its own, however it may be a good bet that something like that is in-plan down the road.
Suffice it to say that all this is complicated stuff and not a subject for the faint of heart. But if you are serious about keeping an eye on one of the most dynamic players in the storage area, the SMARTS technology is certainly one you should track. It, along with the other technologies referenced earlier in this series about EMC (plus EMC’s VMware and its consulting practice) mark EMC as a company readying itself to emerge as a major player across the enterprise.
