Você está na página 1de 6

Web Server Monitoring and

Performance Counter Analysis


Murali Bodagala
Murali.bodagala@tcs.com
TATA Consultancy Services
Hyderabad, India

Abstract
Continuous and end to end server monitoring is require to deliver high available,
performance, reliable and stable web applications. Application performance
depends on various performance counters which needs be carefully monitored on
day to day basis, this article discuss the performance counters that are most
helpful in diagnosing stress and performance related issues in Microsoft ASP.NET
applications, what precautions need to be taken in order to fix the potential
problem, and other resources that can be used to monitor the health of an
ASP.NET application.

Introduction
Monitoring web server performance plays critical role in delivering high scalable
and reliable application, the primary tool for diagnosing a system's performance is
Performance Monitor (PerfMon.exe). Performance Monitor tool collects the
specified performance counters registered by system administrators, these
counters corresponds to a specific aspect of system performance. By selecting the
counters appropriate to the requirement, that helps administrator’s to monitor
ASP.NET Application performance in a better way and can able to diagnose system
performance at a granular level. This paper discusses the important counter sets
that are very useful in monitoring ASP.NET performance.

Performance counters are not just to diagnose performance problems, including


finding memory leaks, but to prevent them from occurring in the first place. In
many organizations, it is common practice to collect the Performance counters in
the organization's servers on a constant basis. It is good practice to set the counter
sampling frequency to low so that Performance data collection minimally impacts
the server's performance. Some organizations collects the sample performance
data is once an hour or a few times a day based on the stability of the system.

Long-term data analysis shows the trend of the increased server load over time for
each server. This information is important for taking preventive actions to
performance related issues that can occure down the road.
The drawback with Performance Monitor is there are hundreds of performance
counters, unless choosing right counters that actually measure the performance
and interpret the collected data rightly, performance counter data is worthless.

Performance counters
Performance counters of the following critical system resources need to be
monitored for web Server and IIS

 Memory
 Processor
 Hard Disk
 Network
 WWW Service
 Active Server Pages

Memory:
Following table contains the counters related to Memory management, these
counters help in determining the problems that are related to memory issues.

Counter Name Ideal value


Pages per second Less than 20 (consider improving RAM if over 80)
Available bytes Maintain over 20MB.
committed bytes Not more than 75 percent of physical memory size
pool non paged bytes Steady (a slow rise may indicate a memory leak)

 Available Memory. Total amount of memory that is left after nonpaged pool
allocations and paged pool allocations
 Pages/sec. This value is indicate number of page inputs and outputs per
second
 Committed Bytes. Total amount of memory that has been allocated for the
exclusive use of any of processes,

 Nonpageable memory pool bytes. This is size of physical system memory


for objects that cannot be written to disk even when they are not being
used. An uncontrolled increase in this value would be indicative of a
memory leak in a Kernel level service or driver.
Processor:
Processor related bottlenecks can be identified by analyzing the following counter
that is listed in below table.

Counter Name Ideal value


% Processor Time Less than 50% on any processor

 % Processor Time is the percentage of elapsed time that the processor


spends to execute a non-Idle thread.

Hard Disk:
Hard disk related bottlenecks can be identified by analyzing the following counters
that is listed in below table.

Counter Name Ideal value


% Disk Time Maintain a low value
PhysicalDisk Avg. Disk Less than 4
Queue Length

 % Disk Time counter monitors the percentage of time that the disk is busy
with read/write activity.
 PhysicalDisk Avg. Disk Queue Length is the average number of both read
and writes requests that were queued for the selected disk during the
sample interval.

Network:
Hard disk related bottlenecks can be identified by analyzing the following counters
that is listed in below table.

Counter Name Ideal value


Bytes Total/sec Less than 70 megabytes/second for 1000
Mbps network adaptor

 Bytes Total/sec indicates the rate at which the network adaptor is


processing data bytes. This counter includes all application and file data.

WWW Service:
WWW service related bottlenecks can be identified by analyzing the following
counters that is listed in below table.

Counter Name Ideal value


Bytes Total/sec As high as possible
File Cache Hits As high as possible
File Cache %, As high as possible
ISAPI Extension No threshold value
Requests/sec

 Bytes Total/sec indicates the rate at which the network adaptor is


processing data bytes. This counter includes all application and file data.
 File Cache Hits is total Number successful lookups for all ASP.NET application
cache calls.
 File Cache % is the ratio of file cache hits to total cache requests. Note that
the Windows counter calculates this value as a historical average since the
last restart, but Load Tester does not.
 ISAPI Extension Requests/sec is the rate that ISAPI Extension requests are
received by the Web service.

WWW Active Server Pages:


The performance counters of the Active Server Pages object should be enabled to
monitor ASP are listed below.

Counter Name Ideal value


Requests/Sec No threshold value
Request Queued maintain a low counter value
Request Wait Time maintain a low counter value

Requests/Sec, The number of requests executed per second, a drop of this counter
during heavy load could mean that your applications might be a bottleneck.

Request Queued The number of ASP.NET requests in the application request


queue.

Request Wait Time is the amount of time the most recent request was waiting in
the queue.

Analysis of Performance Counters


Manually monitoring and analyzing the performance counters is tedious task,
Performance Analysis of Logs (PAL) Tool is makes the analyzing job easy, this tool
analyzes the performance logs and generates the reports with graphs and alters
the administrator’s on the counters that are reaching threshold limits. This tool
can be freely downloaded from www.pal.codeplex.net.

Following screenshots are generated as part of the PAL report by PAL tool for
Physical Disk Avg. Disk Queue Length performance counter; spikes in the graph
represent the Disk Queue Length.
Report also generates the thresholds that were hit for a specific time and their
associated values.

Above report is identified the critical threshold for PhysicalDisk Avg. Disk Queue
Length that could be affecting overall performance, PAL tool allows for very quick
counter analysis to see which counters could be affecting overall server
performance.

Reference

[1]. http://www.tech-faq.com/iis-performance-monitoring-and-tuning.html - IIS


Performance Monitoring and Tuning

[2]. http://msdn.microsoft.com/en-us/library/ms972959.aspx - ASP.NET


Performance Monitoring, and When to Alert Administrators

[3]. http://searchwindowsserver.techtarget.com/tutorial/Performance-
monitoring-tools-for-Windows - Performance monitoring tools for Windows

[4]. http://technet.microsoft.com/en-us/library/cc768048.aspx - Performance


Monitor Counters

[5]. http://pal.codeplex.com/ - Performance Analysis of Logs (PAL) Tool

Você também pode gostar