Windows Reliability and Performance Monitor

Windows Reliability and Performance Monitor

I will let you into a secret, the Reliability and Performance Monitor is a techie’s dream.  This GUI is pure fun to explore, you are certain to find something new and interesting.

Review of Windows Reliability and Performance Monitor

• Why Investigate the Reliability and Performance Monitor?
• Getting Started with the Reliability and Performance Monitor
• Performance Monitor Logs
• What’s New in Windows Performance

Why Investigate the Reliability and Performance Monitor?

The Resource Monitor will give you instant appreciation of which processes are hogging the CPU, Memory, Disk and Network.  When you need a longer history of how processes and applications consume server resources, then investigate the Data Collector Sets.  Begin by using the preset templates, then experiment by modifying the templates and saving your own settings.

If you need to justify the time in ‘playing’ with this tool;  I suggest that the biggest advantage of learning about the Reliability and Performance Monitor is that you will have a sound grasp of the basics when it comes to solving a Windows Server 2008 problem.

Getting Started with the Reliability and Performance Monitor

My favourite way to launch the GUI is to click on the Start (button), type ‘perfmon’ in the ‘Start Search’ box.  When the application launches, I look at the top left of the screen and click on the Reliability and Performance Monitor.  I notice my friend ‘Barking Eddie’ uses a different route, he clicks Start, then selects Administrative Tools.

Another management tactic that I like is creating an MMC, the benefits are that the console remembers your settings, and you can create a whole family of related snap-ins.  If you like this technique then inside the Reliability and Performance Monitor go to the File menu and choose ‘Save as’.

Once you start exploring the GUI, you soon appreciate that the Reliability and Performance Monitor is really 3 utilities in one; the only minor trap is that when you want to work with the Resource Overview you must select ‘Reliability and Performance’ in the left-hand tree.

In a nutshell, this is monitor is intuitive ‘Windows’ at its best; just keep clicking the window pane to see more detail.  At last, an application that lives up to the hype of ‘Easy to use’.  One tip, look out for the up and down arrows circled in the screenshot below. 

Rough and Ready Performance Monitor Troubleshooting

When you want to trace the root cause of a computer running slowly it helps if you ask yourself, and the server, a series of branching questions.  Start with: ‘Is this a hardware or software problem?’  Then follow up with questions such as, ‘Can we rule out a disk failure, a loose SIMM chip or a broken network connection?’

OK, so it’s not a hardware problem.  What you need now is a pincer movement.  To locate the problem ask yourself these two related questions:  ‘Which resource is the bottleneck?’  And, ‘Which program is consuming most of the resource?’  To help identify the program consuming most of the resource I click on the up arrow on the appropriate column, for example, Average CPU, or Commit(KB) in the Memory window.

Here are detailed instruction on how to get started with Performance Monitor

• Begin by clicking high, high, high on precisely Reliability Monitor. (See blue band above).
• Develop a theory – which PROCESS is responsible.  For a quick test, in turn, sort each column with numeric data.  If you are wrong, then what’s your next guess?
• Memory is the most likely bottleneck.  Check that % Used Physical Memory is less than 90%.
• Disk problems could indicate that a disk is about to fail.  Sort on both ‘Read’ and ‘Write’.
• CPU bottlenecks, % continuously over 80%, may indicate this is old server due for replacement.
• Network bottlenecks may need a third-party utility to confirm.  Performance Monitor underplays network problems because it only shows you data from the one machine and not the whole network.

Brutal Advice for Newbies

If a quick check of the Resource Overview does not reveal the source of your problem, AND YOU ARE NEWBIE, then I am sorry but I must be brutal – there is no easy solution.  My best advice is call for an expert, it will be cheaper and less frustrating in the long-run.  OK, so you have no money, but you are willing to spend time trying to isolate the problem.  Internet research is likely to reveal lots people sitting on the fence, the problem is that every setup is different, it’s only experience that tells you if latency is due to an anti-virus program trying to update their image files, and not a SQL database that is suddenly having to service more user queries.

Guy Recommends:  A Free Trial of the Network Performance Monitor (NPM) v12

SolarWinds’ Network Performance Monitor will help you discover what’s happening on your network.  This utility will also guide you through troubleshooting; the dashboard will indicate whether the root cause is a broken link, faulty equipment or resource overload.

Perhaps the NPM’s best feature is the way it suggests solutions to network problems.  Its second best feature is the ability to monitor the health of individual VMware virtual machines.  If you are interested in troubleshooting, and creating network maps, then I recommend that you give this Network Performance Monitor a try.

Download your free trial of SolarWinds Network Performance Monitor.

What’s New in Windows Performance Monitor Compared with Previous Versions of Perfmon

Your old perfmon is still there, but it’s surrounded by useful gadgets for general or level-one troubleshooting, for example, Data Collector Sets.  There are 3 tools which make up the Windows Reliability And Performance Monitor: Resource View, Reliability Monitor, and Performance Monitor (perfmon).  The result is a broader utility which really is more friendly for solving common problems.  Also the System Stability Chart helps you to investigate minor faults before they escalate into a full-scale crisis.

Performance Monitor Logs

The key to understanding the logs is to realize that when you launch the Reliability and Performance Monitor it is set to record live data.  To CREATE logs head for the Data Collector Sets.  To replay logs that made earlier then click on the ‘View Log Data’ icon.  See the two screenshots below.

Data Collector Sets and Templates

If the overview in the Reliability and Performance Monitor does not reveal the root cause the problem, then maybe logging would help.  The Data Collector Sets make stage two of performance monitoring easier, look out for the useful templates with predefined settings, which you can use to generate performance logs. 

The difficulty is that you could be drowned with data.  Through guidance from the Data Collector Sets folder you may be able to record some meaningful data.  However, my guess is that it will take 3 goes to collect data that confirms or denies your theory as to the root cause of your problem.  Whereas the Monitor, really does live up to the hype of easy to use.  All I will say about the Data Collector Templates is that they are EASIER to capture meaningful data than the old perfmon.

Case Studies for Performance Monitoring

Many people believe that you are not supposed to have fun while you learn, but to me, having fun is the ONLY way I learn.  Thus making a game of finding the bottleneck is one of the most effective ways of improving a server’s performance.  Here are possible sources of bottlenecks which cause your server to respond slower than you would expect from its specification.

• A program monopolizes a particular resource.  E.g. Malware hogs the CPU.
• Insufficient resources are available.  E.g. a database consumes all the memory.
• A program, device, or service fails.  E.g. Terminal service times-out.
• Software is incorrectly installed or configured.  E.g. missing, or wrong version of DLL.
• The system is incorrectly configured for the workload.  E.g. not enough memory.

Solutions to Performance Problems

It helps problem solving if you review the range of solution at regular stages.  At one extreme you could buy new faster hardware, or complete system.  Yet at the other extreme, a cost nothing solution such as load balancing could fix the problem at least temporarily.

If you say to yourself, OK, I’ll splash out and by a new system.  Your brain may say, ‘but there is nothing wrong with the disk subsystem’, followed by, ‘Hang on, those quad-processors are already overkill’.  Then a better solution emerges, ‘How about if I just add more memory?’. 

However, an alternative thought process could be, ‘Let’s just by new memory, hmmm… than means that single processor will be the bottleneck.  And 80GB disks looked big 5 years ago, but they are full and these days we can get much bigger disks for the price we paid.  Also, isn’t that fan a bit noisy?  Conclusion, why don’t we buy a balanced system with new components.’

Specific Advice

• Add more resource, e.g. add an extra memory SIMM / DIMM.
• Load-balancing, move some of the processes, or users files, to another disk or even to another server.
• Upgrade, buy a faster disk system.
• Baselines, creating baselines separates the professionals from the amateurs.  It’s just so much easier to solve a problem that arose this week if you know what ‘Normal’ looked like.  When you have baselines from last week, last month and last weekend, then the patterns provide the clues for a speedy resolution.

Summary of Using Performance Monitor

Nothing changes with performance monitoring.  The secret of success remains the same.  Concentrate on which processes are using the big 4 resources: memory, cpu, disk and network.  Look for ‘cost nothing’ solutions, e.g. move the paging file to another disk, perform regular defrags.  Employ the Windows System Resource Manger to meter out resources, and put limits on any users or applications that are bullying the CPU.  If all else fails ask, ‘will installing more RAM speed up a lethargic server?’

Guy Recommends : SolarWinds’ Free VM Monitor

The best feature of this new this new version of SolarWinds VM Monitor is that it checks Windows Hyper-V.  Naturally, it still works with virtual machines on VMware ESX Servers.  VM Monitor is a clever desktop tool that not only tests that your server is online, but also displays the CPU and memory utilization for each node.

It’s easy to install and to configure this virtual machine monitor, all you need the host server’s IP address or hostname and the logon info. Give this virtual machine monitor a try – it’s free.

Download your free copy of SolarWinds VM Monitor.

Reliability Monitor

Vista was the first place that I saw this reliability monitor, at the time I thought, ‘This tool would be even more useful on my servers’.  Well here it as a snap-in for Windows Server 2008. 

The Reliability Monitor is an intelligent filter that trawls the Event Logs and pulls out significant events which it then displays a chart.  Microsoft describe the Reliability Monitor as an intelligent agent for Performance Monitor’s Alerts.

The Reliability snap-in records system stability as a mark out of 10.  Look behind the bare SSI (System Stability Index) number, and research events on the chart to see when software changed, or the services froze.  Observe that the main chart has data lines which record Application, Hardware, Windows (Operating System) failures.  In particular the Reliability Monitor places red crosses on dates when failures occurred.  For any given event, note the detailed description underneath the chart.

Servers can often continue working, albeit slower, even thought there are errors, what this monitor does is show you significant event so that you can decide what corrective action to take.  Replace hardware that’s on the blink, research better drivers, or even consider managing CPU usage with a separate snap-in called the System Resource Monitor.

What you also get is troubleshooters to identify here and now what is preventing the server operating as designed, for example network unavailable, dodgy drivers, loose memory chip.

When calculating the SSI, recent failures are weighted more heavily than past failures, thus once you resolve a problem you should soon see the index rise. 

Tip: To review all your available data, click on the drop-down date menu and select: ‘Select all’.

Remember the other servers in your organization.  As with previous versions of Perfmon, you can collect data from other servers, for example, right-click on ‘Reliability and Performance’ and select: Connect to to another machine.  You really need to select a Windows Server 2008, or a Vista machine, because XP and Windows Server 2003 don’t have the correct agents.

Summary

If you are prepared to put in the time, then using the Reliability and Performance Monitor will reward you with detailed understanding of your Windows Server 2008.  Your explorations will be a labour of love, and the justification of investing time is that you will have the experience and a base-line to make future troubleshooting more successful.

Microsoft Windows Server 2008 Topics:

• Server 2008 Home   •Network Monitor   • Free Syslog Analyser

• AD DC   • Roles   • Features   • Hyper-V   • UAC   • IPv6   • GP Preferences

• SolarWinds Network Performance Monitor   •Performance Monitor