|
|
 |
|
NetIntercept® Case Studies
|
|
 |
Debugging A Distributed Server
Traffic In Context
Network applications are designed to be scalable, responsive and (most of all)
reliable. Thorough testing is a must before an application can go live, but an
application that runs well in the test environment can fail when exposed to the
full stresses of a heavily-loaded network and thousands of users.
Example: Your company has just put its new application into beta. This application
allows customers to log int from around the world and securely access information.
The application does well in its first week out, but with the addition of another
group of users, logins start to fail. How do you find and correct the problem?
The Hard Way
You know the drill: there's a logical sequence of steps you must go through to
make sure nothing has been overlooked. So, your engineers would first go
through the code that performs the login. But a serious error in that section
would have been caught before you took the code to beta; they'll probably be
searching simply to check off that step.
Running the application in a debugger or with multiple log messages enabled, though
another necessary step, isn't likely to find a timing error because the program
isn't running at full capacity.
So you quickly come to the "fun" part: using a network packet analyzer.
Even for the experts, it's painfully slow work to reassemble connections from
from individual packets, and then examining them for anomalies.
And, it could be just one byte out of a million.
NetIntercept's Easy Analysis
NetIntercept reassembles the streams from captured traffic. You get to focus
on the data that's relevant to your application. In a fraction of the time
you thought it would take, you can find out whether a given connection was
properly handled. And, you can readily look at associated connections, such
as those made to databases on other machines.
One client found that it was a matter of scale. Using NetIntercept, they
discovered that when a login failed, it was because a related connection to
a particular database couldn't be completed in time. Further investigation
using different criteria revealed that no matter how many users were logged
in, there were never more than 15 connections to that database.
Within thirty minutes, the limiting header file parameter had been identified
and fixed.
NetIntercept quickly and conveniently collates millions of packets into individual
TCP and UDP connections. And it offers a variety of sophisticated search
criteria, so you can efficiently select and examine the interesting connections.
NetIntercept slashes the time and manpower needed for debugging network applications.
|
|
|
|