Five log data records: allows you to easily analyze system performance

1. slow response time

Response time is the most common and useful performance of log data. It allows you to know how long a request is responded by the system. For example, web server logs can give you an insight into how long it takes for a request to return a response from a client device. This time can include the time when different components (application servers and data blocks) behind the Web server are used to process requests, so it can instantly view how your applications work. The response time recorded by the client device/browser can give you a more comprehensive understanding, because it also captures the page loading and Network Delay Time on the app/browser.

A good rule for measuring response time is the 3 Response Time Limit published by Jakob Nielsen in 1993. This rule is of great significance now. In short, the limit of 0.1 seconds makes users feel like a system instant response. The limit of 1.0 seconds is considered as a constant flow, and the limit of 10 seconds keeps users' attention in conversation.

The slow response time mode almost always follows the following mode:

• Response_time> X

Response_time is the field value, indicating the response of the server or client; "X" is a threshold value. If the threshold value exceeds the field value, your system is very bad for users.

2. Memory problems and garbage collection

The memory overflow error may be disastrous for the system, and the lack of resources may lead to application crash. Therefore, when these events occur, you can create tags and generate alarm notifications.

Memory overflow can be a target direction for garbage collection, which serves as the tracking direction and is notified. The difference between memory insufficiency exceptions and Memory leakage lies in the difference between major system interruptions and simple server restart.

In addition, slow garbage collection may also be the cause of slow user experience. In some cases, garbage collection may slow down and block application operations, knowing that garbage collection is complete.

The following is a common example of memory-related problems:

• Insufficient memory
• Memory limit exceeded
• Memory leakage detection
• Java. Lang. outofmemoryerror
• System. outofmemoryexception
• Memwatch: leak: ended heapdiff
• GC and statistical data

3. deadlock and thread Problems

Deadlocks may occur in many cases and have a bad impact on the system. When a deadlock occurs, it will not stop your system completely but slow down. In short, a deadlock occurs when two competing processes wait for the other party to complete.

Most deadlock patterns only contain the keyword "deadlock", but some common patterns follow the following structure:

• "Deadlock"
• "Deadlock found when trying to get lock"
• "An error occurred while processing the application: deadlock"

4. high resource usage (CPU/Hard Disk/Network)

In many cases, the slowdown in system performance may not be caused by any major software defect. It may be because the load on a simple system increases, but there is no available resources to handle this problem.

Use the following example to analyze the resource usage mode:

• Metric =/cpuutilization/and minimum> X
• CPU> X
• Disk> X
• The disk is in or near its capacity
• Insufficient disk space
• Java. Io. ioexception exception: no space is available on the device.
• Insufficient bandwidth

5. Slow Database Query

This may be useful if you know that the query fails, because it identifies that your request may only have no relevant data when it returns, so as to help you determine that there is no user-required data in the database. However, the more subtle problem is that the user can obtain the correct data, but the result takes a lot of time to return.

Tracking slow queries allows you to track the execution of your database queries. Set an acceptable threshold value for the query time. When any threshold value is exceeded, it helps you identify the user experience that is being affected.

Instance mode:

• Sqlexception
• SQL timeout
• Long query
• Slow Query
• Warning query takes longer than X
• Query_time> X

All in all, using log data can help you identify problems in your system.

Five log data records: allows you to easily analyze system performance