PAL also can provide application-specific analysis for applications such as Microsoft BizTalk Server, Microsoft Exchange Server, Microsoft Office SharePoint Server, Microsoft SQL Server, and Microsoft IIS. So as an administrator wearing several hats, you can have application-specific performance data analyzed without being an expert in the performance counters for an application. PAL can make your life easier by providing analysis for baseline data when performance is typical or to help pinpoint the root cause of a performance issue when a problem occurs.
PAL’s user-friendly UI walks you through the few steps necessary to start the analysis process. The analysis report that PAL generates is an .html file that’s stored by default under the My Documents\PAL Reports folder. The report contains hyperlinks and graphs that enable easy interpretation and navigation, and the file’s portability lets you easily store it in a convenient location.
Using PAL
An example of where PAL saved the day came when a customer received a Microsoft Operations Manager (MOM) alert that detected that all BizTalk services had gone offline on one of the customer’s BizTalk servers. Remotely connecting to the server wasn’t possible then. However, since the server was part of a cluster, the event logs were replicated, revealing the following error at the time the BizTalk server went offline:
Event Type: Error
Event Source: Srv
Event Category: None
Event ID: 2019
Description: The server was unable to allocate from the system non-paged pool because the pool was empty.
The error explained why remote connections to the server weren’t possible but gave us no information about the problem’s root cause.
The customer solved the immediate problem by rebooting the server, which then resumed participating in BizTalk transactions. But we needed to understand what caused the failure. To do so, our first step after the reboot was to capture a Performance Monitor log. After capturing the performance data, we copied the .blg file to our workstation for analysis, then loaded the .blg file into the PAL wizard to generate an analysis report. Using the event log error indicating that the system was depleted of non-paged pool memory, we scrolled down to view the alerts in the report’s memory section and noticed 31 Handle Leak Detection alerts, as Figure 2 shows.
Clicking the Handle Leak Detection hyperlink displayed an easy-to-read graph, which revealed that more than 340,000 handles were consumed by a single process. Additionally, we clicked the Chronological Order link near the top of the report (not shown in Figure 2), which displayed an explanation of the alert, as Figure 3 shows.
The process that PAL alerted us to turned out to be the problem with this particular server after we updated the process (i.e., we checked with the vendor for an update, found an update that the customer had not applied, and applied the update—which solved the problem). Bingo! No more leaking handles, and no more alerts from MOM indicating that BizTalk processes are offline.
Performance Analysis Insight
These tools and the guidance I’ve provided for using them should give you some insight into managing the process of system-performance analysis. This knowledge can help you better understand how your systems typically perform and, if you do encounter a performance problem, help you to resolve it with free and easy-to-use tools.
Special thanks to Kent Weare, a Microsoft MVP who helped contribute to this article.
Register
