68 posts categorized "Paul Offord" Feed

Sharkfest 17 EU - First Thoughts (by Paul Offord)

That's it - another Sharkfest event done and dusted.  Janice and her team did a fantastic job, as always, and the venue was truly spectacular - think of 1930's opulence with a high tech twist and fantastic food.


I was a bit of a late comer - I didn't arrive until midday on the penultimate day - but even so, I could tell it had been a good one.  With attendees from more than 30 countries it was a truly international event.  I was lucky enough to catch several good sessions, and here's a quick summary of some important points I learned.

First up was Kary Rogers.  I think his US session should be compulsory viewing ( see https://youtu.be/tyk2-0MY9p0 ), and the EU session was another tour de force.  He presents straightforward systematic ways of looking at TCP performance, using some Wireshark features I've never thought to use, all mixed in with a dose of comedy.  The EU session had some new stuff and if it becomes available on the Sharkfest Retrospective area, you are going to want to watch it.

Then I sat in on a session called Developer Bytes Lightning Talks–Development Track, hosted by Roland Knall.  There was some real nitty gritty dev stuff in the session, but

Continue reading "Sharkfest 17 EU - First Thoughts (by Paul Offord)" »

Analyzing Microsoft IIS Web Logs - Part 2 (by Paul Offord)

Now almost all the streams we analyze are encrypted, how can we see what's inside those pesky SSL/TLS packets. Here's one way.


In the previous video in this series we saw how web logs provide an abundance of information; just the sort of stuff we need to take a performance problem to a developer.  And now we can analyze web logs with Wireshark.

In this video ...

Continue reading "Analyzing Microsoft IIS Web Logs - Part 2 (by Paul Offord)" »

Analyzing Microsoft IIS Web Logs - Part 1 (by Paul Offord)

Wireshark's new TRANSUM plugin provides a great way to identify slow web site and web service transactions, but there's a problem.  More often than not, web traffic is carried in SSL (TLS) encrypted messages, and so, although we can see slow response times, we can't see the detail.  To prove the cause of a slow response time, ideally we want to see the URI, query strings and, in the case of a web service request, the SOAP Action value.


If we are very lucky, we may be able to get a copy of the private SSL keys and use Wireshark to decrypt the traffic, but what if that's not possible.  The good news is that web logs have much of the information we need, and we can combine this with Wireshark network traces to get a more complete picture.

In this video ...

Continue reading "Analyzing Microsoft IIS Web Logs - Part 1 (by Paul Offord)" »

Palo Alto Packet Latency Case Study Using Workbench and Wireshark (by Paul Offord)

Analyzing packets at two points provides an accurate way to determine the delays across a network.  The team at Advance7 used this technique to find the cause of performance and stability problems with a web application.  The system topology was complex, but very common in today's enterprise environments; users accessing systems using a Windows terminal and ESX VDI-delivered desktops.


Users reported slow response times and intermittent disconnects.  The path through the network from VDI host to application server was 10 GbE all the way, and so link overload was unlikely.  There were various theories about the cause of the problem but solid evidence was needed.

In this video ...

Continue reading "Palo Alto Packet Latency Case Study Using Workbench and Wireshark (by Paul Offord)" »

TCP Checksum Error Case Study (by Paul Offord)

When I see TCP Retransmissions and Dup ACKs in a trace I naturally think about packet loss, but that's not the only cause.  The TCP Checksum mechanism is used to check the integrity of the TCP payload (or segment) and, although it's rare to see genuine checksum errors in a trace, it's another cause of retransmissions.

  Network topology

For Wireshark users there's good and bad news.  The good news is that Wireshark can check each packet for TCP Checksum errors.  The bad news is that they are not always genuine errors.  So how can we tell the difference?

In this video ...

Continue reading "TCP Checksum Error Case Study (by Paul Offord)" »

Using NetworkMiner with a Windows netsh trace File (by Paul Offord)

Before analyzing a network packet trace file, I try to make sure that I've collected information about IP addresses and TCP/UDP port numbers.  Even so, I still find that I don't have all the information I need.  There are techniques you can use to get the missing information - check NBNS host announcements, explore the names resolved by DNS - but it's all just more hassle.


Recently I noticed a bit of a buzz around NetworkMiner, so I thought I'd check it out.  What I found was a simple tool that does just what I need; extract useful host and service information from Wireshark traces.  We now analyze a fair number of traces captured with Windows netsh trace, so I thought I'd look at how we can use NetworkMiner with these Windows-native trace files.

In this video ...

Continue reading "Using NetworkMiner with a Windows netsh trace File (by Paul Offord)" »