성능 문제 진단

Redis is known for its incredible speed, but even the fastest systems can face performance issues under certain conditions. Keeping your application responsive means understanding how to diagnose these slowdowns.

In this lesson, we'll learn to identify common Redis bottlenecks and how to investigate them.

When Redis performance dips, it's often due to one of these three main culprits:

• Slow Commands: Individual operations taking too long to execute.
• High Memory Usage: Redis consuming excessive RAM, leading to eviction or swapping.
• Network Latency: Delays in communication between your application and the Redis server.

Let's dive into how to spot each of these.

Redis has a built-in Slow Log that records commands whose execution time exceeds a specified threshold. This is extremely valuable for pinpointing problematic operations.

First, let's check the current threshold:

The `slowlog-log-slower-than` value is in microseconds (1,000,000 microseconds = 1 second). By default, it's often set to 10,000 (10 milliseconds).

To retrieve the latest slow commands, use `SLOWLOG GET`:

Each entry in the Slow Log provides key details:

• Unique ID: An identifier for the log entry.
• Timestamp: When the command was executed.
• Duration: How long the command took in microseconds.
• Command: The actual command and its arguments.

Look for commands with high durations, as these are the ones slowing down your Redis instance.

Redis stores all its data in RAM. If Redis consumes too much memory, it can lead to severe performance issues:

• Eviction: Redis might start deleting keys to free up space, losing data.
• Swapping: The operating system might move Redis data to disk, drastically slowing down all operations.
• Crashes: In extreme cases, Redis might crash due to out-of-memory errors.

The `INFO memory` command is your primary tool for monitoring Redis's memory consumption. It provides detailed statistics, including how much memory Redis is currently using.

Run this command in your Redis CLI:

When you run `INFO memory`, pay close attention to these fields:

• used_memory_human: The total memory consumed by Redis, in a human-readable format (e.g., 1.50M, 10.2G).
• used_memory_peak_human: The maximum memory Redis has ever used since starting.
• mem_fragmentation_ratio: The ratio of physical memory used to logical memory requested. A high ratio (>1.5) can indicate memory fragmentation.

Even if Redis commands are fast and memory usage is optimal, network delays can make your application feel slow. Network latency is the time it takes for data to travel between your client and the Redis server.

High latency adds overhead to every Redis operation, impacting overall performance.

To check for network issues:

• Ping Test: Use the `PING` command from your client. A slow response indicates network latency.
• Location: Ensure your Redis client application is geographically close to your Redis server. Distant servers mean higher latency.
• Network Path: Investigate firewalls, VPNs, or cloud network configurations that might be introducing delays.

You suspect your Redis instance is experiencing performance issues due to slow commands. Which Redis CLI command is specifically designed to help you identify these?

Great job! You've learned how to identify and begin diagnosing common Redis performance issues:

• Use `SLOWLOG GET` to find long-running commands.
• Check `INFO memory` and `MEMORY USAGE` for memory-related problems.
• Consider network proximity and `PING` to check for latency.

These tools are crucial for keeping your Redis instance running smoothly!