Bootsnap

Beta-quality. See the last section of this README.

Bootsnap is a library that plugs into Ruby, with optional support for ActiveSupport and YAML, to optimize and cache expensive computations. See How Does This Work.

Performance

• Discourse reports a boot time reduction of approximately 50%, from roughly 6 to 3 seconds on one machine;
• One of our smaller internal apps also sees a reduction of 50%, from 3.6 to 1.8 seconds;
• The core Shopify platform -- a rather large monolithic application -- boots about 75% faster, dropping from around 25s to 6.5s.

Usage

This gem works on MacOS and Linux.

Add bootsnap to your Gemfile:

gem 'bootsnap'

If you are using rails, add this to config/boot.rb immediately after require 'bundler/setup':

require 'bootsnap/setup'

If you are not using rails, or if you are but want more control over things, add this to your application setup immediately after require 'bundler/setup' (i.e. as early as possible: the sooner this is loaded, the sooner it can start optimizing things)

require 'bootsnap'
env = ENV['RAILS_ENV'] || "development"
Bootsnap.setup(
  cache_dir:            'tmp/cache',          # Path to your cache
  development_mode:     env == 'development', # Current working environment, e.g. RACK_ENV, RAILS_ENV, etc
  load_path_cache:      true,                 # Optimize the LOAD_PATH with a cache
  autoload_paths_cache: true,                 # Optimize ActiveSupport autoloads with cache
  disable_trace:        true,                 # (Alpha) Set `RubyVM::InstructionSequence.compile_option = { trace_instruction: false }`
  compile_cache_iseq:   true,                 # Compile Ruby code into ISeq cache, breaks coverage reporting.
  compile_cache_yaml:   true                  # Compile YAML into a cache
)

Protip: You can replace require 'bootsnap' with BootLib::Require.from_gem('bootsnap', 'bootsnap') using this trick. This will help optimize boot time further if you have an extremely large $LOAD_PATH.

How does this work?

Bootsnap optimizes methods to cache results of expensive computations, and can be grouped into two broad categories:

• Path Pre-Scanning
  • Kernel#require and Kernel#load are modified to eliminate $LOAD_PATH scans.
  • ActiveSupport::Dependencies.{autoloadable_module?,load_missing_constant,depend_on} are overridden to eliminate scans of ActiveSupport::Dependencies.autoload_paths.
• Compilation caching
  • RubyVM::InstructionSequence.load_iseq is implemented to cache the result of ruby bytecode compilation.
  • YAML.load_file is modified to cache the result of loading a YAML object in MessagePack format (or Marshal, if the message uses types unsupported by MessagePack).

Path Pre-Scanning

(This work is a minor evolution of bootscale).

Upon initialization of bootsnap or modification of the path (e.g. $LOAD_PATH), Bootsnap::LoadPathCache will fetch a list of requirable entries from a cache, or, if necessary, perform a full scan and cache the result.

Later, when we run (e.g.) require 'foo', ruby would iterate through every item on our $LOAD_PATH ['x', 'y', ...], looking for x/foo.rb, y/foo.rb, and so on. Bootsnap instead looks at all the cached requirables for each $LOAD_PATH entry and substitutes the full expanded path of the match ruby would have eventually chosen.

If you look at the syscalls generated by this behaviour, the net effect is that what would previously look like this:

open  x/foo.rb # (fail)
# (imagine this with 500 $LOAD_PATH entries instead of two)
open  y/foo.rb # (success)
close y/foo.rb
open  y/foo.rb
...

becomes this:

open y/foo.rb
...

Exactly the same strategy is employed for methods that traverse ActiveSupport::Dependencies.autoload_paths if the autoload_paths_cache option is given to Bootsnap.setup.

The following diagram flowcharts the overrides that make the *_path_cache features work.

Bootsnap classifies path entries into two categories: stable and volatile. Volatile entries are scanned each time the application boots, and their caches are only valid for 30 seconds. Stable entries do not expire -- once their contents has been scanned, it is assumed to never change.

The only directories considered "stable" are things under the Ruby install prefix (RbConfig::CONFIG['prefix'], e.g. /usr/local/ruby or ~/.rubies/x.y.z), and things under the Gem.path (e.g. ~/.gem/ruby/x.y.z) or Bundler.bundle_path. Everything else is considered "volatile".

In addition to the Bootsnap::LoadPathCache::Cache source, this diagram may help clarify how entry resolution works:

It's also important to note how expensive LoadErrors can be. If ruby invokes require 'something', but that file isn't on $LOAD_PATH, it takes 2 * $LOAD_PATH.length filesystem accesses to determine that. Bootsnap caches this result too, raising a LoadError without touching the filesystem at all.

Compilation Caching

(A more readable implementation of this concept can be found in yomikomu).

Ruby has complex grammar and parsing it is not a particularly cheap operation. Since 1.9, Ruby has translated ruby source to an internal bytecode format, which is then executed by the Ruby VM. Since 2.2, Ruby exposes an API that allows caching that bytecode. This allows us to bypass the relatively-expensive compilation step on subsequent loads of the same file.

We also noticed that we spend a lot of time loading YAML documents during our application boot, and that MessagePack and Marshal are much faster at deserialization than YAML, even with a fast implementation. We use the same strategy of compilation caching for YAML documents, with the equivalent of Ruby's "bytecode" format being a MessagePack document (or, in the case of YAML documents with types unsupported by MessagePack, a Marshal stream).

These compilation results are stored in a cache directory, with filenames generated by taking a hash of the full expanded path of the input file (FNV1a-64).

Whereas before, the sequence of syscalls generated to require a file would look like:

open    /c/foo.rb -> m
fstat64 m
close   m
open    /c/foo.rb -> o
fstat64 o
fstat64 o
read    o
read    o
...
close   o

With bootsnap, we get:

open      /c/foo.rb -> n
fstat64   n
close     n
open      /c/foo.rb -> n
fstat64   n
open      (cache) -> m
read      m
read      m
close     m
close     n

This may look worse at a glance, but underlies a large performance difference.

(The first three syscalls in both listings -- open, fstat64, close -- are not inherently useful. This ruby patch optimizes them out when coupled with bootsnap.)

Bootsnap writes a cache file containing a 64 byte header followed by the cache contents. The header is a cache key including several fields:

• version, hardcoded in bootsnap. Essentially a schema version;
• os_version, A hash of the current kernel version (on macOS, BSD) or glibc version (on Linux);
• compile_option, which changes with RubyVM::InstructionSequence.compile_option does;
• ruby_revision, the version of Ruby this was compiled with;
• size, the size of the source file;
• mtime, the last-modification timestamp of the source file when it was compiled; and
• data_size, the number of bytes following the header, which we need to read it into a buffer.

If the key is valid, the result is loaded from the value. Otherwise, it is regenerated and clobbers the current cache.

Putting it all together

Imagine we have this file structure:

/
├── a
├── b
└── c
    └── foo.rb

And this $LOAD_PATH:

["/a", "/b", "/c"]

When we call require 'foo' without bootsnap, Ruby would generate this sequence of syscalls:

open    /a/foo.rb -> -1
open    /b/foo.rb -> -1
open    /c/foo.rb -> n
close   n
open    /c/foo.rb -> m
fstat64 m
close   m
open    /c/foo.rb -> o
fstat64 o
fstat64 o
read    o
read    o
...
close   o

With bootsnap, we get:

open      /c/foo.rb -> n
fstat64   n
close     n
open      /c/foo.rb -> n
fstat64   n
open      (cache) -> m
read      m
read      m
close     m
close     n

If we call require 'nope' without bootsnap, we get:

open    /a/nope.rb -> -1
open    /b/nope.rb -> -1
open    /c/nope.rb -> -1
open    /a/nope.bundle -> -1
open    /b/nope.bundle -> -1
open    /c/nope.bundle -> -1

...and if we call require 'nope' with bootsnap, we get...

# (nothing!)

Trustworthiness

We use the *_path_cache features in production and haven't experienced any issues in a long time.

The compile_cache_* features work well for us in development on macOS. It should work on Linux, and we intend to deploy it in production, but we haven't yet.

disable_trace should be completely safe, but we don't really use it because some people like to use tools that make use of trace instructions.

feature	where we're using it
load_path_cache	everywhere
autoload_path_cache	everywhere
disable_trace	nowhere, but it's safe unless you need tracing
compile_cache_iseq	development, but probably safe to use everywhere
compile_cache_yaml	development, but probably safe to use everywhere