It covers EVERYTHING you need:

✔ What PerfMon is
✔ Why PerfMon is used
✔ When to use PerfMon
✔ PerfMon Architecture
✔ Key Terminologies
✔ How PerfMon works
✔ How to use PerfMon step-by-step
✔ All important counters
✔ Real performance engineering use cases
✔ Advantages, Limitations, Best Practices
✔ Troubleshooting with PerfMon
✔ PerfMon vs other tools

---

📘 PERFMON (Performance Monitor): The Complete One-Stop Tutorial

---

1. Introduction to PerfMon

What is PerfMon?

PerfMon (Performance Monitor) is a built-in Windows performance monitoring tool used to collect, visualize, and analyze system and application performance metrics called performance counters.

It allows you to monitor:

• CPU usage
• Memory usage
• Disk I/O
• Network traffic
• Process-level metrics
• Thread, handle, and heap usage
• .NET CLR health (GC, exceptions, JIT, threads)
• IIS / ASP.NET metrics
• SQL Server counters
• Custom application counters

PerfMon is one of the most powerful and most underutilized diagnostic tools in Windows.

---

2. Why PerfMon Exists (Purpose)

PerfMon helps you:

✔ Detect performance bottlenecks

CPU spikes, memory leaks, disk contention, network congestion

✔ Troubleshoot production issues

• High CPU
• Low memory
• Disk thrashing
• Network bottleneck
• Slow API response time
• GC pauses
• SQL performance issues

✔ Perform load testing diagnostics

It allows you to correlate load tests with system metrics.

✔ Monitor .NET applications

GC behavior, exceptions/sec, thread pool usage

✔ Monitor IIS / Kestrel

Requests/sec, connection queue, request pipeline health

✔ Monitor SQL Server

Batch requests/sec, compilations, cache hits

PerfMon is essential for Performance Engineering, SRE, DevOps, and Developers.

---

3. When to Use PerfMon (Scenarios)

Use PerfMon when:

1️⃣ You run a load test

You must capture perf counters to see:

• CPU
• Memory
• Disk
• Network
• .NET GC
• SQL Server

2️⃣ Production troubleshooting

When a server is slow, PerfMon reveals patterns.

3️⃣ Memory leaks

Watch Private Bytes + GC counters.

4️⃣ CPU spikes

Check Processor Time + Context Switches + thread count.

5️⃣ Slow disk or file reads

Disk Queue Length is key.

6️⃣ Slow API throughput

Monitor Requests/sec + ASP.NET pipeline limits.

7️⃣ High GC pauses

% Time in GC tells the story.

8️⃣ SQL Server performance issues

Batch Requests/sec + Compilations/sec.

PerfMon is used by senior engineers, SREs, DevOps, DBAs, and Performance Test Engineers.

---

4. PerfMon Architecture

PerfMon works on a simple architecture:

 Windows Kernel + Applications
         |
   Performance Counters (APIs)
         |
     PerfMon.exe
         |
   - Live Graphs
   - Reports
   - Data Collector Sets

Key points:

• Windows exposes more than 5,000 counters.
• PerfMon reads these counters and shows them.
• Counters can come from OS, .NET CLR, SQL Server, IIS, third-party apps.

---

5. Key Terminology

➤ Performance Counter

A real-time metric exposed by Windows or applications.

Examples:

• % Processor Time
• Available Memory
• Disk Queue Length
• Bytes Received/sec
• .NET CLR Memory: % Time in GC
• SQL Compilations/sec

---

➤ Object

A category of counters, e.g.,

• Processor
• Memory
• PhysicalDisk
• Network Interface
• Process
• .NET CLR Memory

---

➤ Instance

A specific process or core.

Examples:

• Processor → _Total, 0, 1, 2
• Process → dotnet, chrome, myapp.exe
• SQL Server → MSSQLSERVER

---

➤ Data Collector Set (DCS)

A group of counters scheduled to log into a file.

Used for:

• Load tests
• Production monitoring
• Historical analysis

---

➤ BLG / CSV Logs

PerfMon logs stored for later analysis.

BLG is the binary format (smaller, efficient).

---

6. PerfMon Components

PerfMon has three major sections:

---

6.1 Performance Monitor (Live Graph)

Shows real-time counters as a line graph.

Great for:

• Spikes
• Bottleneck identification
• Correlation during load tests

---

6.2 Data Collector Sets

You can create a DCS that:

• Runs for hours/days
• Captures hundreds of counters
• Saves in BLG or CSV

Used in:

• Load testing
• Production issues
• Long-running performance tests

---

6.3 Reports

After a DCS completes, PerfMon auto-generates reports summarizing:

• Peak CPU
• Memory trend
• Disk latency
• Network usage

---

7. How to Use PerfMon (Step-by-Step Tutorial)

---

Step 1: Launch PerfMon

Press:

Win + R → perfmon

OR
Search: Performance Monitor

---

Step 2: Open “Performance Monitor”

Left panel → Monitoring Tools → Performance Monitor

You will see:

• A default red CPU line
• Live graph

---

Step 3: Add Counters

Click:

Green + icon → Add Counters

You will see categories like:

• Processor
• Memory
• Disk
• Network
• Process
• .NET CLR Memory
• SQLServer:*

---

Step 4: Select Important Counters

Choose your object → select instance → click Add.

Examples:

For CPU

Processor → % Processor Time → _Total
Process → % Processor Time → dotnet

For Memory

Memory → Available MBytes
Memory → % Committed Bytes
Process → Private Bytes → myapp.exe
Code language: CSS (css)

For Disk

PhysicalDisk → Avg. Disk Queue Length → _Total

For Network

Network Interface → Bytes Total/sec
Code language: PHP (php)

For .NET

.NET CLR Memory → % Time in GC → myapp.exe
.NET CLR Exceptions → # Exceptions/sec → myapp.exe
Code language: PHP (php)

---

Step 5: Analyze Graphs

Watch the graphs for:

• Spikes
• Trends
• Degradation
• Correlation with load

PerfMon lets you:

• Change line colors
• Change scale
• Pause
• Zoom in
• Pin counters

---

Step 6: Create Data Collector Set (DCS)

Why?

For long-term monitoring (load test, 1 hour+).

Steps:

1. Right-click → Data Collector Sets → User Defined
2. New → Data Collector Set
3. Add performance counters
4. Set sample interval (1 sec or 5 sec)
5. Choose log path
6. Click Finish
7. Start the DCS

You now have automatic background logging.

---

Step 7: View Reports

After DCS completes:

Reports → User Defined → Your DCS

You will see:

• CPU peaks
• Memory trend
• Disk I/O charts
• Network summary

---

8. Most Important PerfMon Counters

Here are the mission-critical counters for troubleshooting.

---

CPU

Counter	Why
Processor → % Processor Time	CPU bottleneck (>80%)
Process → % Processor Time	Per-process CPU
System → Context Switches/sec	Thread contention
Processor → % Privileged Time	Kernel overhead

---

Memory

Counter	Why
Memory → Available MBytes	Memory pressure
% Committed Bytes In Use	If >80% → close to paging
Process → Private Bytes	Memory leak indicator
Paging File → % Usage	Disk paging

---

Disk

Counter	Why
Avg. Disk Queue Length	Disk bottleneck if >2 per spindle
Avg. sec/Read	Disk read latency
Avg. sec/Write	Disk write latency

---

Network

Counter	Why
Bytes Total/sec	Throughput
Output Queue Length	NIC congestion
TCP Retransmissions/sec	Network issues

---

.NET CLR

Counter	Why
% Time in GC	GC pauses
# Gen 2 Collections	Full GCs are expensive
# Exceptions/sec	Exception storms
LOH Size	LOH fragmentation
Thread Count	Thread pool starvation

---

ASP.NET / IIS

Counter	Why
Requests/sec	Throughput
Request Execution Time	API delay
Current Connections	Server load
Pipeline Instance Count	Bottlenecks

---

SQL Server

Counter	Why
Batch Requests/sec	Query throughput
Compilations/sec	Poor plan caching
Buffer Cache Hit Ratio	Memory pressure
Page Life Expectancy	SQL memory issues

---

9. PerfMon Use Cases

---

✔ Use Case 1: High CPU

Counters:

• Processor → % Processor Time
• Process → % Processor Time
• Context Switches/sec
• Thread Count

You identify:

• Hot processes
• Bad loops
• Thread contention

---

✔ Use Case 2: Memory Leak

Monitor:

• Private Bytes
• Working Set
• % Committed Bytes
• GC counters

---

✔ Use Case 3: Disk Bottleneck

Monitor:

• Disk Queue Length
• Disk latency (Avg sec/Read)
• Page File usage

---

✔ Use Case 4: Slow API Response

Monitor:

• ASP.NET Request Time
• Requests/sec
• % Time in GC
• SQL Batch Requests/sec

---

10. Advantages of PerfMon

✔ Free – built into Windows

✔ Extremely lightweight

✔ Logs run for days

✔ Can monitor local + remote servers

✔ Thousands of counters

✔ Full historical reporting

✔ Used in enterprise load testing

✔ Works with Power BI, Excel, Grafana, Splunk

---

11. Limitations of PerfMon

❌ Not beginner-friendly

❌ No anomaly detection

❌ Only Windows

❌ Requires manual setup

❌ BLG analysis requires external tools

---

12. Best Practices

✔ Always create a DCS for load tests
✔ Use 1-sec sampling for short tests
✔ Use 5-sec sampling for long tests
✔ Always include CPU, Memory, Disk, Network
✔ For .NET apps → add CLR counters
✔ For SQL tests → add SQL counters
✔ Log to BLG (not CSV)
✔ Use naming like “Test1_CPU_Memory_Disk”

---

13. PerfMon vs Other Tools

Tool	Purpose
PerfMon	Deep OS + .NET diagnostics
Resource Monitor	Real-time troubleshooting
Task Manager	Basic monitoring
Process Explorer	Process-level details
dotnet-counters	.NET runtime live metrics
dotnet-trace	Trace collection
SQL Profiler	SQL tracing

---

14. Conclusion

PerfMon is one of the most essential performance tools for:

• Developers
• SRE
• DevOps
• Performance Test Engineers
• System Admins

It gives complete visibility into:

• System health
• .NET runtime
• SQL Server
• Disk I/O
• Network
• Application behavior

No Windows performance investigation is complete without PerfMon.

---

Just tell me.