This is definitely a belief held by system administrators, since their comfort zone is perfectly aligned with such a problem. Slow app means slow network means add more pipe. Easy peasy.
But in actuality, there is evidence to suggest that the network is actually the problem far less times than most people are aware. For example, what about the network application that shoves a huge amount of data over the network and back to the server, only to have the receiving server utilize just 10 or 15 percent of the data, dumping the rest? Through one lens, you might be tempted to think that this problem could be solved by a larger pipe. But in reality, this is an application fix: re-tool the app to send back just what the data center needs, and suddenly you’ve reduced inbound traffic from the app by up to 90 percent.
There are operational issues that can look like network issues, too. Like backing up data on the same network infrastructure on which your application is running. These sorts of things are just asking for trouble, especially when your business depends on the most efficient transactions possible.
This is why Application Performance Management solutions that put a lot of emphasis on the performance on the network should be taken with a grain of salt. Yes, the network should be examined too, but it is not the be-all-end-all of Application Performance Management. End user experience monitoring and business transaction monitoring need to be combined with end-to-end monitoring to give a more holistic view of what’s going on with the application.
Sure, if you find a misconfigured router somewhere on your network and that gives you a boost in speed once it’s fixed, that’s great. But such easy solutions are not always going to be there, and even if they are, who’s to say that’s the best your application can do? What about the interface? And the data being sent back-and-forth over the network?
Application Performance Management is not about just looking at one piece of the puzzle, fixing it, and declaring it done. That’s like fixing someone’s broken leg after a car accident and declaring them healed–all the while missing that concussion they got.
The reason why the specific approaches are still prevalent in Application Performance Management is that–unlike a real accident victim–most applications aren’t that broken. Many apps are usually really close to where they need to be, so finding room for improvement must involved looking at all aspects of the application, not just this or that component.
By all means, check your network out, but make sure it’s not all you check. Even if you score the big fix, you may still have work to do.