rdb

The rdb command is a tool for parsing Redis RDB files (supporting versions 1 through 12, up to Redis 7.2). It provides various commands for memory analysis, data conversion, and key searching.

Run rdb without arguments to see the full usage manual.

The rdb tool provides several analysis modes:

• Memory Report: Generate a CSV memory report using -c memory.
• AOF Conversion: Convert RDB to an AOF file using -c aof.
• Big Keys: Find the largest keys using -c bigkey. Use -n to limit the number of results.
• Hot Keys: Find the most frequently used keys via LFU frequency using -c hotkey. This requires the Redis instance to have been configured with maxmemory-policy allkeys-lfu or volatile-lfu.
• Prefix Analysis: Analyze memory usage by key prefixes using -c prefix. Use -prefix-sep to specify a separator for constant memory analysis.

rdb -c memory -o memory.csv dump.rdb
rdb -c aof -o dump.aof dump.rdb
rdb -c bigkey -n 10 dump.rdb
rdb -c hotkey -n 50 dump.rdb

You can install the rdb tool using go install if you have a Go environment set up, or by downloading the executable from the releases page and adding it to your PATH.

go install github.com/hdt3213/rdb@latest

You can generate a FlameGraph to analyze which types of keys occupy the most memory. This starts a web service on a specified port.

Use the -c flamegraph command. You can specify the port with -port and the key separator with -sep.

rdb -c flamegraph -port 16379 -sep : dump.rdb

Convert a Redis RDB file into a JSON array of key-value objects. You can control the concurrency of the conversion using the -concurrent flag.

Use -show-global-meta to include Redis metadata (like version and creation time) and Lua function definitions in the output.

# Basic conversion
rdb -c json -o dump.json dump.rdb

# Conversion with 8 concurrent workers
rdb -c json -o intset_16.json -concurrent 8 cases/intset_16.rdb

# Conversion including global metadata
rdb -c json -o function.json -show-global-meta cases/function.rdb

Use the bigkey command to find the largest keys in the RDB file. Use the hotkey command to find the most frequently accessed keys based on LFU (Least Frequently Used) frequency.

Note: hotkey analysis requires that the original Redis instance was configured with maxmemory-policy allkeys-lfu or volatile-lfu.

# Get top 10 largest keys
rdb -c bigkey -n 10 dump.rdb

# Get top 50 hottest keys
rdb -c hotkey -n 50 -o hotkey.csv dump.rdb

You can install the rdb tool using Go, Homebrew, or by downloading pre-built binaries.

Using Go

If you have Go installed, run:

go install github.com/hdt3213/rdb@latest

Using Homebrew

brew install rdb

Using Binaries

Download the executable binary from the releases page and add it to your PATH.

Generate a flamegraph to visualize which types of keys are consuming the most memory. This starts a web service on the specified port.

You can use the -sep flag to define how keys should be split for the flamegraph visualization.

rdb -c flamegraph -port 16379 -sep : dump.rdb

To convert a Redis RDB file to a JSON format, use the json command. You can optionally specify the output path and the number of concurrent converters to speed up the process.

To include global metadata (like redis-ver, ctime, used-mem, and functions), add the -show-global-meta flag.

rdb -c json -o dump.json dump.rdb

# With concurrency and metadata
rdb -c json -o function.json -show-global-meta cases/function.rdb

Generate a CSV report estimating the memory usage of each key in the RDB file based on its encoded size. The report includes columns for database index, key, type, size (in bytes), human-readable size, and element count.

rdb -c memory -o mem.csv cases/memory.rdb

Analyze the distribution of keys by their prefixes. This is useful for understanding key naming patterns and their impact on memory.

• Standard mode: Uses a radix tree for analysis.
• Constant memory mode: Use the -prefix-sep flag to specify a separator. This uses separator-based analysis which is more memory-efficient for large datasets.

# Standard prefix analysis (top 10, max depth 3)
rdb -c prefix -n 10 -max-depth 3 -o prefix-report.csv dump.rdb

# Constant memory prefix analysis using ':' as separator
rdb -c prefix -n 10 -max-depth 3 -prefix-sep : -o prefix-report.csv dump.rdb

Prefix analysis (-c prefix) helps understand memory distribution by key prefixes.

Key Options:

• -n <number>: Number of results to return.
• -max-depth <int>: Maximum depth of the prefix tree.
• -o <path>: Output file path (e.g., for a CSV report).
• -prefix-sep <separator>: Use this flag to enable flat-map mode (constant memory). Instead of a radix tree, it uses the specified separator to perform analysis. You can provide multiple separators by repeating the flag.

Example: To perform prefix analysis using : as a separator in constant memory mode: rdb -c prefix -n 10 -max-depth 3 -prefix-sep : -o prefix-report.csv dump.rdb