The path to gaining visibility and insight into an application and its underlying network has been a long standing requirement for enterprises. As a result, there have been a number of players in the application performance management space and their corresponding market size has grown considerably over the years. It’s not rocket science – knowing what’s happening to your applications carries both operational as well as top line business value. After all, you need to know when your system isn’t working. Even more importantly, you need to know when there is a way you can improve the end user experience so they can get more business done.

Historically, this has been achieved using one of three ways: (1) Aggregate logs from application servers, NetFlow data from routers, and syslog data from infrastructure; (2) Use software agents to collect data from the application server directly; (3) Use network taps to collect data about applications transactions that include relevant network performance data. The first two give good data and enable a wide variety of analysis. The last is arguably the most powerful as it effectively ties the application details down to the end user experience in a directly measurable way.

The problem with the last method is that the footprint from which the information is collected is rapidly losing ground. In order to put network taps onto the wire, you have to have control of the network and a place to put them. As an increasing number of applications go virtual and the use of cloud (both public and private) emerge, these network footprints to place the tap are altogether disappearing. After all, could you imagine the conversation where we ask a cloud service provider to let us drop a network tap so we can see all their packets?

To solve this problem, we have to look to other places where the information lives but doesn’t require a physical change to the network -- and that place is the existing network infrastructure.

Consider for a moment what something like a NetScaler knows about a given transaction: It knows the network experience because the TCP/IP stack needs to track very detailed timing information in order to optimize what packets it sends; it knows what the application is doing because it is tracking the app state for the purpose of load balancing; finally, it knows the server health and performance based on its network. All of that information, put in the hands of the right analytics tool, is immensely valuable. 

But what about the physical footprint of the NetScaler? I can’t very well ask my cloud provider to use MY hardware in the cloud for the same reason I can’t ask them to put my tap on their network. Furthermore, which analytics tool makes the most sense? Depending on the need, types of reports, and other kinds of analysis that can be done, different tools will appeal to different organizations. 

The first matter is easy – the NetScaler is available in a virtual form. In fact, any modern infrastructure product that can be virtualized is being virtualized. That removes the issue of physical presence. The latter is a more subtle problem.

When we started work on AppFlow, we thought long and hard about this. We could come up with a way of getting deep application insight without a tap, but if no one was on the other side to take the data and turn it into something interesting, the data would be useless. Partnering with a few analytics companies is always possible, but we were bound to miss a group of users that need another kind of analytics tool. What we really wanted to do was make it possible for anyone in the analytics space to get the data and run with it. 

The result? We opened up the AppFlow specification and decoupled ourselves from it. In essence, the data had become democratized -- anyone that wanted to work with the data and make it useful could do so. Almost immediately we saw players from the analytics space jump on this and add support. Some we knew, a few we didn’t. Even more significantly, NetScaler no longer was the only piece of infrastructure that could generate this data. Any device that holds application state information (e.g., firewalls and WAN Optimization devices) can generate this data and share a different perspective on the network. The more perspectives that are shared with analytics tools, the more powerful the resulting analysis becomes.

From our customer’s perspective, the interest has been overwhelming. Already, at least one customer has put an order for network taps valued > $1M on hold as they look to see how they can leverage their existing network footprints. A few phone calls and their analytics provider is on-board to support AppFlow. 

So as you think about what value you want to get from your existing network footprint, be sure to think about AppFlow. All the data you need, in a democratized form, that anyone and everyone can process and turn into business intelligence. You can read more about AppFlow at www.appflow.org.