* The rise of application flow management
Enterprise Management Associates is finally wrapping up our definition of a taxonomy for application flow management to follow a report on NetFlow adoption.
Application flow management, as we define it, includes pair-based flow management such as that in RMON and NetFlow, as well as protocol and packet analysis – from Layers 2 through 7 and beyond, potentially. The common definer is a focus on “bloodstream” vs. component or device-centric management.
EMA deliberately took this broader view because we’re seeing the virtually hundreds of vendors in this market begin to cross over into network and application performance management, service-level management and service discovery, capacity planning, optimization, application prioritization, accounting and chargeback, route analytics, VoIP and security. Application flows are enabling a whole host of disciplines, and the technology is evolving far beyond traditional handheld packet analyzers at one end, and grand but expensive RMON investments on the other. In effect, this class of application-flow vendors is emerging from being quietly interesting, to becoming one of the most innovative areas within enterprise and service provider management.
Here’s why:
* Capturing application flows is more real-time than polling, and often more contextual. Many of these products can provide good “first warnings” to activate and focus device-centric polling.
* Awareness of traffic volume can be key in planning and optimizing infrastructure for application behavior – whether it’s for capacity planning, traffic engineering, accounting and chargeback, or QoS and application prioritization.
* Understanding anomalous behavior can be valuable in troubleshooting and in security. The flipside of “anomalous” is “normal,” and understanding what’s normal may be challenging since it often varies by time of day or day of the week, but it can also be enormously useful in performance management and “on-demand” strategies.
* Capturing application traffic flows in terms of consumption patterns reveals a lot about the infrastructure, but it can also reveal a lot about the human beings consuming the IT services in terms of both their appetites and their disaffections. Once you can summarize insights such as “cost,” “quality” and “hunger,” you’ve gone a long way toward getting the information you need to really take control of your IT service requirements.
Cost and deployment requirements vary almost as widely as functionality across the application flow management marketplace. Products may be probe- or appliance-centric, or they may leverage network instrumentation such as NetFlow and sFlow. Placements of agents, probes, or collectors may be centralized around the data center, may be at the WAN edge, or may require localized attention in many branch locations.
Charting your way through the jungle of hundreds of offerings will take you from low-cost entrants (several thousand dollars and less) to deployments that begin well into the six figures. Integration strategies with other management software vary greatly by vendor – and there are of course many good reasons why you should favor application flow products with a rich sense of industry “citizenship” over those that are more focused as stand-alone or niche values. All this richness is good for you, the buyer, if you’re willing to take some time in shopping and planning your requirements.
In the end, there are no generic right or wrong answers – the right answers are what will fit your particular requirements the best. EMA’s goal has been to develop something of a roadmap (or perhaps to keep the analogy more accurate, a “path map”) through this fertile wilderness of hundreds of vendors.
