Every big network seems to have an impossible problem. It's something that defies all the normal failure modes and baffles everyone who examines it. The combination of networks, IT systems, applications, and UC systems that keep businesses running today create a very complex puzzle, with many interacting components. We are ultimately left with fragile systems in which small changes create unexpected problems.
Fortunately, there are often reasonable explanations for the behaviors exhibited by networks and the systems they support. At NetCraftsmen, we are frequently posed a problem that the customer has been enduring for a long time, only to have one of our consultants solve it in a few hours or days. The depth and breadth of experience means that we've frequently seen the problem before or have a good enough understanding of the fundamental technology to find a good explanation for it.
Tools, Data, and Analysis
There are times when there simply isn't enough data about a problem to offer an explanation and a solution. This is when tools should be used to collect the data that provides the basis of a good analysis. On more than one occasion, I have used a long-running ping operation to identify network problems. It is useful to plot the ping round trip times and look for periodic anomalies. Use a small bit of editing or scripting to create a CSV file to import into a spreadsheet and plot the resulting ping round trip times on one axis and the packet count on the other axis. This method has been particularly useful for identifying routing and periodic network congestion problems.
In another case, I had to use packet captures to collect enough data to identify the source of a problem. To explain, a customer wanted to transfer large CAD files over 1Gbps WAN links. The transfers should have taken 10 seconds, but were taking much, much longer, making their remote CAD software useless. It took a week of detailed WireShark (the tool) analysis of several packet captures (the data) and some additional reading about the fundamental operation of TCP in some specific operating systems (the analysis) to identify the cause of the slow performance. In this case, we were ultimately able to pinpoint the problem and make several recommendations for working around the file server's operating system limitations.
Do You Have an Impossible Problem?
Bring your impossible problems to Enterprise Connect 2018, for the session Ask the Expert: An Interactive Session on Network & Systems Management, taking place on March 14 at 1:30 p.m. ET.
Having presented at Enterprise Connect for many years now, I have been asked a number of interesting questions in past sessions. For this year, I worked with conference organizers to provide a place where you can ask a panel of experts about your problems directly. Our panelists are senior consultants who have seen a lot of things across the spectrum of IT systems, including wireless, networking, network management, and unified communications. Of course, we can't get into detailed solutions during the limited time of the session, but we can provide you with some ideas on approaches to solving the problems you're facing.
At a minimum, you should be able to leave the session with a list of next steps and perhaps some new ideas of things to check. And if you want to give the experts a jump on thinking about your particular issue, you can send your questions in advance to [email protected].
If we can't answer your question, we should at least be able to get you started down some new paths or refer you to someone who can help. Join us at Enterprise Connect for this lively session.
And learn more about Systems Management & Network Design at Enterprise Connect 2018, March 12 to 15, in Orlando, Fla. Register now using the code NOJITTER to save an additional $200 off the Early Bird Pricing or get a free Expo Plus pass.
Related content: