Performance Series | EP10, Network Congestion. How IT Pros (SHOULD) Troubleshoot Slow PC’s and Servers

In episode 10 we’ll talk about network congestion. While the problem isn’t normally on your server, I’ll show you how to determine what caused the issue if it is. This episode also talks about how to spot and identify network latency. We cover perfmon and how to interpret the counters, pathping, TCPView, IO Meter as a network tool, and a handful of other tips and tricks.

Performance Series | EP9, Processor Bottlenecks. How IT Pros (SHOULD) Troubleshoot Slow PC’s and Servers

Episode 9, Processor Bottlenecks. Performance Series.

In episode 9 we’ll load a machine up to nearly 100% processor utilization and talk about how to track it in perfmon (because I know you already know how to do this with task manager). We also cover context switching, DPC’s, Interrupts, and the Processor Queue Length. After this we cover the process counters and thresholds for tracking down whodunn’it.

Performance Series | EP8, Kernel Memory Leaks. How IT Pros (SHOULD) Troubleshoot Slow PC’s and Servers

In episode 8 we’ll cover how to identify and troubleshoot memory leaks in the kernel virtual address space. I’ll demo a system crash/lockup and then show what it looks like in perfmon, then expand on how to identify the offending pooltag in the kernel using poolmon.exe. We specifically focus on the nonpaged pool, the paged pool, and system PTE’s as well as the pertinent counters that correlate with thresholds of each “bucket”.

Performance Series | EP7, Windows Memory Architecture. How IT Pros (SHOULD) Troubleshoot Slow PC’s and Servers

Episode 7, Windows Memory Architecture. Performance Series.

In episode 7 we’ll start talking about “Fake Memory” as I’ve been eluding to for several episodes. We cover the microkernel memory architecture of 32 and 64 bit operating systems, the Virtual Memory Manager, and why user mode processes sometimes crash when they exceed it. We cover this in a whiteboarding session and then move into a demo of an actual application crash, and show what a typical crash looks like in Perfmon.

Performance Series | EP5, Tracking RAM and Page File Exhaustion: How IT Pros (SHOULD) Troubleshoot Slow PC’s and Servers

Episode 5, Tracking RAM and Page File Exhaustion. Performance Series.

In episode 5 we start the discussion of memory leaks. This is a topic that will span several issues. Memory can mean the virtual address space of a process or the kernel itself, or it could mean you ran out or RAM. Or, you ran out of RAM and Page File. How do you tell which one? How do you figure out which process stole all your precious memory. Specifically in this episode we’ll be talking about what it looks like when you run out of both RAM and Page File (the most common type of process memory leak). We also cover the difference between committed bytes of a process verses private bytes. Finally, we cover concepts like the Virtual Memory Manager of Windows. Trimming of the working set. What the page file really is, etc. The next episode will continue with a discussion more related to the working set.

Performance Series | Episode 4, Troubleshooting Disk Bottlenecks. Performance Series.

In episode 4 we cover the first of the actual counter sets used to identify problems, and the methodology to do so. Identifying a disk bottleneck involves more than just latency counters, it also includes ruling out whether your server or PC was actually the culprit. So what we will do in this episode is load up a computer with a lot of disk activity, then additionally load up the hypervisor that runs that computer and compare the differences. Finally, we will do some process analysis and determine who the bad guy was.

Performance Series | EP2, Perfmon Unleashed: How IT Pros Troubleshoot Slow PC’s and Servers (or, how they SHOULD)

In episode 2 we continue working with some of the more advanced features of Perfmon, such as setting up data collector sets, managing Perfmon through the command line with logman, and integration with task scheduler.