leveldb Documentation

Technical Documentation: google/leveldb

ℹ️ Provenance: Multi-Tier Ingestion · CodeWiki Reference

Overview (`README.md`) Copy Section

LevelDB is a fast key-value storage library written at Google that provides an ordered mapping from string keys to string values.

This repository is receiving very limited maintenance. We will only review the following types of changes.

• Fixes for critical bugs, such as data loss or memory corruption
• Changes absolutely needed by internally supported leveldb clients. These typically fix breakage introduced by a language/standard library/OS update

Authors: Sanjay Ghemawat ([email protected]) and Jeff Dean ([email protected])

Features

• Keys and values are arbitrary byte arrays.
• Data is stored sorted by key.
• Callers can provide a custom comparison function to override the sort order.
• The basic operations are Put(key,value), Get(key), Delete(key).
• Multiple changes can be made in one atomic batch.
• Users can create a transient snapshot to get a consistent view of data.
• Forward and backward iteration is supported over the data.
• Data is automatically compressed using the Snappy compression library, but Zstd compression is also supported.
• External activity (file system operations etc.) is relayed through a virtual interface so users can customize the operating system interactions.

Documentation

LevelDB library documentation is online and bundled with the source code.

Limitations

• This is not a SQL database. It does not have a relational data model, it does not support SQL queries, and it has no support for indexes.
• Only a single process (possibly multi-threaded) can access a particular database at a time.
• There is no client-server support builtin to the library. An application that needs such support will have to wrap their own server around the library.

Getting the Source

git clone --recurse-submodules https://github.com/google/leveldb.git

Building

This project supports CMake out of the box.

Build for POSIX Copy Section

Quick start:

mkdir -p build && cd build
cmake -DCMAKE_BUILD_TYPE=Release .. && cmake --build .

Building for Windows Copy Section

First generate the Visual Studio 2017 project/solution files:

mkdir build
cd build
cmake -G "Visual Studio 15" ..

The default default will build for x86. For 64-bit run:

cmake -G "Visual Studio 15 Win64" ..

To compile the Windows solution from the command-line:

devenv /build Debug leveldb.sln

or open leveldb.sln in Visual Studio and build from within.

Please see the CMake documentation and CMakeLists.txt for more advanced usage.

Contributing to the leveldb Project

This repository is receiving very limited maintenance. We will only review the following types of changes.

• Bug fixes
• Changes absolutely needed by internally supported leveldb clients. These typically fix breakage introduced by a language/standard library/OS update

The leveldb project welcomes contributions. leveldb's primary goal is to be
a reliable and fast key/value store. Changes that are in line with the
features/limitations outlined above, and meet the requirements below,
will be considered.

Contribution requirements:

1. Tested platforms only. We generally will only accept changes for
   platforms that are compiled and tested. This means POSIX (for Linux and
   macOS) or Windows. Very small changes will sometimes be accepted, but
   consider that more of an exception than the rule.

2. Stable API. We strive very hard to maintain a stable API. Changes that
   require changes for projects using leveldb might be rejected without
   sufficient benefit to the project.

3. Tests: All changes must be accompanied by a new (or changed) test, or
   a sufficient explanation as to why a new (or changed) test is not required.

4. Consistent Style: This project conforms to the
   Google C++ Style Guide.
   To ensure your changes are properly formatted please run:

   text Copy

   clang-format -i --style=file <file>

We are unlikely to accept contributions to the build configuration files, such
as CMakeLists.txt. We are focused on maintaining a build configuration that
allows us to test that the project works in a few supported configurations
inside Google. We are not currently interested in supporting other requirements,
such as different operating systems, compilers, or build systems.

Submitting a Pull Request Copy Chapter

Before any pull request will be accepted the author must first sign a
Contributor License Agreement (CLA) at https://cla.developers.google.com/.

In order to keep the commit timeline linear
squash
your changes down to a single commit and rebase
on google/leveldb/main. This keeps the commit timeline linear and more easily sync'ed
with the internal repository at Google. More information at GitHub's
About Git rebase page.

Performance

Here is a performance report (with explanations) from the run of the
included db_bench program. The results are somewhat noisy, but should
be enough to get a ballpark performance estimate.

Setup Copy Chapter

We use a database with a million entries. Each entry has a 16 byte
key, and a 100 byte value. Values used by the benchmark compress to
about half their original size.

LevelDB:    version 1.1
Date:       Sun May  1 12:11:26 2011
CPU:        4 x Intel(R) Core(TM)2 Quad CPU    Q6600  @ 2.40GHz
CPUCache:   4096 KB
Keys:       16 bytes each
Values:     100 bytes each (50 bytes after compression)
Entries:    1000000
Raw Size:   110.6 MB (estimated)
File Size:  62.9 MB (estimated)

Write performance Copy Chapter

The "fill" benchmarks create a brand new database, in either
sequential, or random order. The "fillsync" benchmark flushes data
from the operating system to the disk after every operation; the other
write operations leave the data sitting in the operating system buffer
cache for a while. The "overwrite" benchmark does random writes that
update existing keys in the database.

fillseq      :       1.765 micros/op;   62.7 MB/s
fillsync     :     268.409 micros/op;    0.4 MB/s (10000 ops)
fillrandom   :       2.460 micros/op;   45.0 MB/s
overwrite    :       2.380 micros/op;   46.5 MB/s

Each "op" above corresponds to a write of a single key/value pair.
I.e., a random write benchmark goes at approximately 400,000 writes per second.

Each "fillsync" operation costs much less (0.3 millisecond)
than a disk seek (typically 10 milliseconds). We suspect that this is
because the hard disk itself is buffering the update in its memory and
responding before the data has been written to the platter. This may
or may not be safe based on whether or not the hard disk has enough
power to save its memory in the event of a power failure.

Read performance Copy Chapter

We list the performance of reading sequentially in both the forward
and reverse direction, and also the performance of a random lookup.
Note that the database created by the benchmark is quite small.
Therefore the report characterizes the performance of leveldb when the
working set fits in memory. The cost of reading a piece of data that
is not present in the operating system buffer cache will be dominated
by the one or two disk seeks needed to fetch the data from disk.
Write performance will be mostly unaffected by whether or not the
working set fits in memory.

readrandom  : 16.677 micros/op;  (approximately 60,000 reads per second)
readseq     :  0.476 micros/op;  232.3 MB/s
readreverse :  0.724 micros/op;  152.9 MB/s

LevelDB compacts its underlying storage data in the background to
improve read performance. The results listed above were done
immediately after a lot of random writes. The results after
compactions (which are usually triggered automatically) are better.

readrandom  : 11.602 micros/op;  (approximately 85,000 reads per second)
readseq     :  0.423 micros/op;  261.8 MB/s
readreverse :  0.663 micros/op;  166.9 MB/s

Some of the high cost of reads comes from repeated decompression of blocks
read from disk. If we supply enough cache to the leveldb so it can hold the
uncompressed blocks in memory, the read performance improves again:

readrandom  : 9.775 micros/op;  (approximately 100,000 reads per second before compaction)
readrandom  : 5.215 micros/op;  (approximately 190,000 reads per second after compaction)

Repository contents Copy Chapter

See doc/index.md for more explanation. See
doc/impl.md for a brief overview of the implementation.

The public interface is in include/leveldb/*.h. Callers should not include or
rely on the details of any other header files in this package. Those
internal APIs may be changed without warning.

Guide to header files:

• include/leveldb/db.h: Main interface to the DB: Start here.

• include/leveldb/options.h: Control over the behavior of an entire database,
  and also control over the behavior of individual reads and writes.

• include/leveldb/comparator.h: Abstraction for user-specified comparison function.
  If you want just bytewise comparison of keys, you can use the default
  comparator, but clients can write their own comparator implementations if they
  want custom ordering (e.g. to handle different character encodings, etc.).

• include/leveldb/iterator.h: Interface for iterating over data. You can get
  an iterator from a DB object.

• include/leveldb/write_batch.h: Interface for atomically applying multiple
  updates to a database.

• include/leveldb/slice.h: A simple module for maintaining a pointer and a
  length into some other byte array.

• include/leveldb/status.h: Status is returned from many of the public interfaces
  and is used to report success and various kinds of errors.

• include/leveldb/env.h:
  Abstraction of the OS environment. A posix implementation of this interface is
  in util/env_posix.cc.

• include/leveldb/table.h, include/leveldb/table_builder.h: Lower-level modules that most
  clients probably won't use directly.

Overview (`port/README.md`) Copy Section

This directory contains interfaces and implementations that isolate the
rest of the package from platform details.

Code in the rest of the package includes "port.h" from this directory.
"port.h" in turn includes a platform specific "port_.h" file
that provides the platform specific implementation.

See port_stdcxx.h for an example of what must be provided in a platform
specific header file.

Doc Index (`doc/index.md`) Copy Section

leveldb

Jeff Dean, Sanjay Ghemawat

The leveldb library provides a persistent key value store. Keys and values are
arbitrary byte arrays. The keys are ordered within the key value store
according to a user-specified comparator function.

Opening A Database Copy Chapter

A leveldb database has a name which corresponds to a file system directory. All
of the contents of database are stored in this directory. The following example
shows how to open a database, creating it if necessary:

#include <cassert>
#include "leveldb/db.h"

leveldb::DB* db;
leveldb::Options options;
options.create_if_missing = true;
leveldb::Status status = leveldb::DB::Open(options, "/tmp/testdb", &db);
assert(status.ok());
...

If you want to raise an error if the database already exists, add the following
line before the leveldb::DB::Open call:

options.error_if_exists = true;

Status Copy Chapter

You may have noticed the leveldb::Status type above. Values of this type are
returned by most functions in leveldb that may encounter an error. You can check
if such a result is ok, and also print an associated error message:

leveldb::Status s = ...;
if (!s.ok()) cerr << s.ToString() << endl;

Closing A Database Copy Chapter

When you are done with a database, just delete the database object. Example:

... open the db as described above ...
... do something with db ...
delete db;

Reads And Writes Copy Chapter

The database provides Put, Delete, and Get methods to modify/query the database.
For example, the following code moves the value stored under key1 to key2.

std::string value;
leveldb::Status s = db->Get(leveldb::ReadOptions(), key1, &value);
if (s.ok()) s = db->Put(leveldb::WriteOptions(), key2, value);
if (s.ok()) s = db->Delete(leveldb::WriteOptions(), key1);

Atomic Updates Copy Chapter

Note that if the process dies after the Put of key2 but before the delete of
key1, the same value may be left stored under multiple keys. Such problems can
be avoided by using the WriteBatch class to atomically apply a set of updates:

#include "leveldb/write_batch.h"
...
std::string value;
leveldb::Status s = db->Get(leveldb::ReadOptions(), key1, &value);
if (s.ok()) {
  leveldb::WriteBatch batch;
  batch.Delete(key1);
  batch.Put(key2, value);
  s = db->Write(leveldb::WriteOptions(), &batch);
}

The WriteBatch holds a sequence of edits to be made to the database, and these
edits within the batch are applied in order. Note that we called Delete before
Put so that if key1 is identical to key2, we do not end up erroneously dropping
the value entirely.

Apart from its atomicity benefits, WriteBatch may also be used to speed up
bulk updates by placing lots of individual mutations into the same batch.

Synchronous Writes Copy Chapter

By default, each write to leveldb is asynchronous: it returns after pushing the
write from the process into the operating system. The transfer from operating
system memory to the underlying persistent storage happens asynchronously. The
sync flag can be turned on for a particular write to make the write operation
not return until the data being written has been pushed all the way to
persistent storage. (On Posix systems, this is implemented by calling either
fsync(...) or fdatasync(...) or msync(..., MS_SYNC) before the write
operation returns.)

leveldb::WriteOptions write_options;
write_options.sync = true;
db->Put(write_options, ...);

Asynchronous writes are often more than a thousand times as fast as synchronous
writes. The downside of asynchronous writes is that a crash of the machine may
cause the last few updates to be lost. Note that a crash of just the writing
process (i.e., not a reboot) will not cause any loss since even when sync is
false, an update is pushed from the process memory into the operating system
before it is considered done.

Asynchronous writes can often be used safely. For example, when loading a large
amount of data into the database you can handle lost updates by restarting the
bulk load after a crash. A hybrid scheme is also possible where every Nth write
is synchronous, and in the event of a crash, the bulk load is restarted just
after the last synchronous write finished by the previous run. (The synchronous
write can update a marker that describes where to restart on a crash.)

WriteBatch provides an alternative to asynchronous writes. Multiple updates
may be placed in the same WriteBatch and applied together using a synchronous
write (i.e., write_options.sync is set to true). The extra cost of the
synchronous write will be amortized across all of the writes in the batch.

Concurrency Copy Chapter

A database may only be opened by one process at a time. The leveldb
implementation acquires a lock from the operating system to prevent misuse.
Within a single process, the same leveldb::DB object may be safely shared by
multiple concurrent threads. I.e., different threads may write into or fetch
iterators or call Get on the same database without any external synchronization
(the leveldb implementation will automatically do the required synchronization).
However other objects (like Iterator and WriteBatch) may require external
synchronization. If two threads share such an object, they must protect access
to it using their own locking protocol. More details are available in the public
header files.

Iteration Copy Chapter

The following example demonstrates how to print all key,value pairs in a
database.

leveldb::Iterator* it = db->NewIterator(leveldb::ReadOptions());
for (it->SeekToFirst(); it->Valid(); it->Next()) {
  cout << it->key().ToString() << ": "  << it->value().ToString() << endl;
}
assert(it->status().ok());  // Check for any errors found during the scan
delete it;

The following variation shows how to process just the keys in the range
[start,limit):

for (it->Seek(start);
   it->Valid() && it->key().ToString() < limit;
   it->Next()) {
  ...
}

You can also process entries in reverse order. (Caveat: reverse iteration may be
somewhat slower than forward iteration.)

for (it->SeekToLast(); it->Valid(); it->Prev()) {
  ...
}

Snapshots Copy Chapter

Snapshots provide consistent read-only views over the entire state of the
key-value store. ReadOptions::snapshot may be non-NULL to indicate that a
read should operate on a particular version of the DB state. If
ReadOptions::snapshot is NULL, the read will operate on an implicit snapshot
of the current state.

Snapshots are created by the DB::GetSnapshot() method:

leveldb::ReadOptions options;
options.snapshot = db->GetSnapshot();
... apply some updates to db ...
leveldb::Iterator* iter = db->NewIterator(options);
... read using iter to view the state when the snapshot was created ...
delete iter;
db->ReleaseSnapshot(options.snapshot);

Note that when a snapshot is no longer needed, it should be released using the
DB::ReleaseSnapshot interface. This allows the implementation to get rid of
state that was being maintained just to support reading as of that snapshot.

Slice Copy Chapter

The return value of the it->key() and it->value() calls above are instances
of the leveldb::Slice type. Slice is a simple structure that contains a length
and a pointer to an external byte array. Returning a Slice is a cheaper
alternative to returning a std::string since we do not need to copy
potentially large keys and values. In addition, leveldb methods do not return
null-terminated C-style strings since leveldb keys and values are allowed to
contain '\0' bytes.

C++ strings and null-terminated C-style strings can be easily converted to a
Slice:

leveldb::Slice s1 = "hello";

std::string str("world");
leveldb::Slice s2 = str;

A Slice can be easily converted back to a C++ string:

std::string str = s1.ToString();
assert(str == std::string("hello"));

Be careful when using Slices since it is up to the caller to ensure that the
external byte array into which the Slice points remains live while the Slice is
in use. For example, the following is buggy:

leveldb::Slice slice;
if (...) {
  std::string str = ...;
  slice = str;
}
Use(slice);

When the if statement goes out of scope, str will be destroyed and the backing
storage for slice will disappear.

Comparators Copy Chapter

The preceding examples used the default ordering function for key, which orders
bytes lexicographically. You can however supply a custom comparator when opening
a database. For example, suppose each database key consists of two numbers and
we should sort by the first number, breaking ties by the second number. First,
define a proper subclass of leveldb::Comparator that expresses these rules:

class TwoPartComparator : public leveldb::Comparator {
 public:
  // Three-way comparison function:
  //   if a < b: negative result
  //   if a > b: positive result
  //   else: zero result
  int Compare(const leveldb::Slice& a, const leveldb::Slice& b) const {
    int a1, a2, b1, b2;
    ParseKey(a, &a1, &a2);
    ParseKey(b, &b1, &b2);
    if (a1 < b1) return -1;
    if (a1 > b1) return +1;
    if (a2 < b2) return -1;
    if (a2 > b2) return +1;
    return 0;
  }

  // Ignore the following methods for now:
  const char* Name() const { return "TwoPartComparator"; }
  void FindShortestSeparator(std::string*, const leveldb::Slice&) const {}
  void FindShortSuccessor(std::string*) const {}
};

Now create a database using this custom comparator:

TwoPartComparator cmp;
leveldb::DB* db;
leveldb::Options options;
options.create_if_missing = true;
options.comparator = &cmp;
leveldb::Status status = leveldb::DB::Open(options, "/tmp/testdb", &db);
...

Backwards compatibility Copy Section

The result of the comparator's Name method is attached to the database when it
is created, and is checked on every subsequent database open. If the name
changes, the leveldb::DB::Open call will fail. Therefore, change the name if
and only if the new key format and comparison function are incompatible with
existing databases, and it is ok to discard the contents of all existing
databases.

You can however still gradually evolve your key format over time with a little
bit of pre-planning. For example, you could store a version number at the end of
each key (one byte should suffice for most uses). When you wish to switch to a
new key format (e.g., adding an optional third part to the keys processed by
TwoPartComparator), (a) keep the same comparator name (b) increment the
version number for new keys (c) change the comparator function so it uses the
version numbers found in the keys to decide how to interpret them.

Performance Copy Chapter

Performance can be tuned by changing the default values of the types defined in
include/options.h.

Block size Copy Section

leveldb groups adjacent keys together into the same block and such a block is
the unit of transfer to and from persistent storage. The default block size is
approximately 4096 uncompressed bytes. Applications that mostly do bulk scans
over the contents of the database may wish to increase this size. Applications
that do a lot of point reads of small values may wish to switch to a smaller
block size if performance measurements indicate an improvement. There isn't much
benefit in using blocks smaller than one kilobyte, or larger than a few
megabytes. Also note that compression will be more effective with larger block
sizes.

Compression Copy Section

Each block is individually compressed before being written to persistent
storage. Compression is on by default since the default compression method is
very fast, and is automatically disabled for uncompressible data. In rare cases,
applications may want to disable compression entirely, but should only do so if
benchmarks show a performance improvement:

leveldb::Options options;
options.compression = leveldb::kNoCompression;
... leveldb::DB::Open(options, name, ...) ....

Cache Copy Section

The contents of the database are stored in a set of files in the filesystem and
each file stores a sequence of compressed blocks. If options.block_cache is
non-NULL, it is used to cache frequently used uncompressed block contents.

#include "leveldb/cache.h"

leveldb::Options options;
options.block_cache = leveldb::NewLRUCache(100 * 1048576);  // 100MB cache
leveldb::DB* db;
leveldb::DB::Open(options, name, &db);
... use the db ...
delete db
delete options.block_cache;

Note that the cache holds uncompressed data, and therefore it should be sized
according to application level data sizes, without any reduction from
compression. (Caching of compressed blocks is left to the operating system
buffer cache, or any custom Env implementation provided by the client.)

When performing a bulk read, the application may wish to disable caching so that
the data processed by the bulk read does not end up displacing most of the
cached contents. A per-iterator option can be used to achieve this:

leveldb::ReadOptions options;
options.fill_cache = false;
leveldb::Iterator* it = db->NewIterator(options);
for (it->SeekToFirst(); it->Valid(); it->Next()) {
  ...
}
delete it;

Key Layout Copy Section

Note that the unit of disk transfer and caching is a block. Adjacent keys
(according to the database sort order) will usually be placed in the same block.
Therefore the application can improve its performance by placing keys that are
accessed together near each other and placing infrequently used keys in a
separate region of the key space.

For example, suppose we are implementing a simple file system on top of leveldb.
The types of entries we might wish to store are:

filename -> permission-bits, length, list of file_block_ids
file_block_id -> data

We might want to prefix filename keys with one letter (say '/') and the
file_block_id keys with a different letter (say '0') so that scans over just
the metadata do not force us to fetch and cache bulky file contents.

Filters Copy Section

Because of the way leveldb data is organized on disk, a single Get() call may
involve multiple reads from disk. The optional FilterPolicy mechanism can be
used to reduce the number of disk reads substantially.

leveldb::Options options;
options.filter_policy = NewBloomFilterPolicy(10);
leveldb::DB* db;
leveldb::DB::Open(options, "/tmp/testdb", &db);
... use the database ...
delete db;
delete options.filter_policy;

The preceding code associates a Bloom filter based filtering policy with the
database. Bloom filter based filtering relies on keeping some number of bits of
data in memory per key (in this case 10 bits per key since that is the argument
we passed to NewBloomFilterPolicy). This filter will reduce the number of
unnecessary disk reads needed for Get() calls by a factor of approximately
a 100. Increasing the bits per key will lead to a larger reduction at the cost
of more memory usage. We recommend that applications whose working set does not
fit in memory and that do a lot of random reads set a filter policy.

If you are using a custom comparator, you should ensure that the filter policy
you are using is compatible with your comparator. For example, consider a
comparator that ignores trailing spaces when comparing keys.
NewBloomFilterPolicy must not be used with such a comparator. Instead, the
application should provide a custom filter policy that also ignores trailing
spaces. For example:

class CustomFilterPolicy : public leveldb::FilterPolicy {
 private:
  leveldb::FilterPolicy* builtin_policy_;

 public:
  CustomFilterPolicy() : builtin_policy_(leveldb::NewBloomFilterPolicy(10)) {}
  ~CustomFilterPolicy() { delete builtin_policy_; }

  const char* Name() const { return "IgnoreTrailingSpacesFilter"; }

  void CreateFilter(const leveldb::Slice* keys, int n, std::string* dst) const {
    // Use builtin bloom filter code after removing trailing spaces
    std::vector<leveldb::Slice> trimmed(n);
    for (int i = 0; i < n; i++) {
      trimmed[i] = RemoveTrailingSpaces(keys[i]);
    }
    builtin_policy_->CreateFilter(trimmed.data(), n, dst);
  }
};

Advanced applications may provide a filter policy that does not use a bloom
filter but uses some other mechanism for summarizing a set of keys. See
leveldb/filter_policy.h for detail.

Checksums Copy Chapter

leveldb associates checksums with all data it stores in the file system. There
are two separate controls provided over how aggressively these checksums are
verified:

ReadOptions::verify_checksums may be set to true to force checksum
verification of all data that is read from the file system on behalf of a
particular read. By default, no such verification is done.

Options::paranoid_checks may be set to true before opening a database to make
the database implementation raise an error as soon as it detects an internal
corruption. Depending on which portion of the database has been corrupted, the
error may be raised when the database is opened, or later by another database
operation. By default, paranoid checking is off so that the database can be used
even if parts of its persistent storage have been corrupted.

If a database is corrupted (perhaps it cannot be opened when paranoid checking
is turned on), the leveldb::RepairDB function may be used to recover as much
of the data as possible

Approximate Sizes Copy Chapter

The GetApproximateSizes method can used to get the approximate number of bytes
of file system space used by one or more key ranges.

leveldb::Range ranges[2];
ranges[0] = leveldb::Range("a", "c");
ranges[1] = leveldb::Range("x", "z");
uint64_t sizes[2];
db->GetApproximateSizes(ranges, 2, sizes);

The preceding call will set sizes[0] to the approximate number of bytes of
file system space used by the key range [a..c) and sizes[1] to the
approximate number of bytes used by the key range [x..z).

Environment Copy Chapter

All file operations (and other operating system calls) issued by the leveldb
implementation are routed through a leveldb::Env object. Sophisticated clients
may wish to provide their own Env implementation to get better control.
For example, an application may introduce artificial delays in the file IO
paths to limit the impact of leveldb on other activities in the system.

class SlowEnv : public leveldb::Env {
  ... implementation of the Env interface ...
};

SlowEnv env;
leveldb::Options options;
options.env = &env;
Status s = leveldb::DB::Open(options, ...);

Porting Copy Chapter

leveldb may be ported to a new platform by providing platform specific
implementations of the types/methods/functions exported by
leveldb/port/port.h. See leveldb/port/port_example.h for more details.

In addition, the new platform may need a new default leveldb::Env
implementation. See leveldb/util/env_posix.h for an example.

Other Information Copy Chapter

Details about the leveldb implementation may be found in the following
documents:

1. Implementation notes
2. Format of an immutable Table file
3. Format of a log file

Doc Impl (`doc/impl.md`) Copy Section

Files Copy Chapter

The implementation of leveldb is similar in spirit to the representation of a
single Bigtable tablet (section 5.3).
However the organization of the files that make up the representation is
somewhat different and is explained below.

Each database is represented by a set of files stored in a directory. There are
several different types of files as documented below:

Log files Copy Section

A log file (*.log) stores a sequence of recent updates. Each update is appended
to the current log file. When the log file reaches a pre-determined size
(approximately 4MB by default), it is converted to a sorted table (see below)
and a new log file is created for future updates.

A copy of the current log file is kept in an in-memory structure (the
memtable). This copy is consulted on every read so that read operations
reflect all logged updates.

Sorted tables Copy Chapter

A sorted table (*.ldb) stores a sequence of entries sorted by key. Each entry is
either a value for the key, or a deletion marker for the key. (Deletion markers
are kept around to hide obsolete values present in older sorted tables).

The set of sorted tables are organized into a sequence of levels. The sorted
table generated from a log file is placed in a special young level (also
called level-0). When the number of young files exceeds a certain threshold
(currently four), all of the young files are merged together with all of the
overlapping level-1 files to produce a sequence of new level-1 files (we create
a new level-1 file for every 2MB of data.)

Files in the young level may contain overlapping keys. However files in other
levels have distinct non-overlapping key ranges. Consider level number L where
L >= 1. When the combined size of files in level-L exceeds (10^L) MB (i.e., 10MB
for level-1, 100MB for level-2, ...), one file in level-L, and all of the
overlapping files in level-(L+1) are merged to form a set of new files for
level-(L+1). These merges have the effect of gradually migrating new updates
from the young level to the largest level using only bulk reads and writes
(i.e., minimizing expensive seeks).

Manifest Copy Section

A MANIFEST file lists the set of sorted tables that make up each level, the
corresponding key ranges, and other important metadata. A new MANIFEST file
(with a new number embedded in the file name) is created whenever the database
is reopened. The MANIFEST file is formatted as a log, and changes made to the
serving state (as files are added or removed) are appended to this log.

Current Copy Section

CURRENT is a simple text file that contains the name of the latest MANIFEST
file.

Info logs Copy Section

Informational messages are printed to files named LOG and LOG.old.

Others Copy Section

Other files used for miscellaneous purposes may also be present (LOCK, *.dbtmp).

Level 0 Copy Chapter

When the log file grows above a certain size (4MB by default):
Create a brand new memtable and log file and direct future updates here.

In the background:

1. Write the contents of the previous memtable to an sstable.
2. Discard the memtable.
3. Delete the old log file and the old memtable.
4. Add the new sstable to the young (level-0) level.

Compactions Copy Chapter

When the size of level L exceeds its limit, we compact it in a background
thread. The compaction picks a file from level L and all overlapping files from
the next level L+1. Note that if a level-L file overlaps only part of a
level-(L+1) file, the entire file at level-(L+1) is used as an input to the
compaction and will be discarded after the compaction. Aside: because level-0
is special (files in it may overlap each other), we treat compactions from
level-0 to level-1 specially: a level-0 compaction may pick more than one
level-0 file in case some of these files overlap each other.

A compaction merges the contents of the picked files to produce a sequence of
level-(L+1) files. We switch to producing a new level-(L+1) file after the
current output file has reached the target file size (2MB). We also switch to a
new output file when the key range of the current output file has grown enough
to overlap more than ten level-(L+2) files. This last rule ensures that a later
compaction of a level-(L+1) file will not pick up too much data from
level-(L+2).

The old files are discarded and the new files are added to the serving state.

Compactions for a particular level rotate through the key space. In more detail,
for each level L, we remember the ending key of the last compaction at level L.
The next compaction for level L will pick the first file that starts after this
key (wrapping around to the beginning of the key space if there is no such
file).

Compactions drop overwritten values. They also drop deletion markers if there
are no higher numbered levels that contain a file whose range overlaps the
current key.

Timing Copy Section

Level-0 compactions will read up to four 1MB files from level-0, and at worst
all the level-1 files (10MB). I.e., we will read 14MB and write 14MB.

Other than the special level-0 compactions, we will pick one 2MB file from level
L. In the worst case, this will overlap ~ 12 files from level L+1 (10 because
level-(L+1) is ten times the size of level-L, and another two at the boundaries
since the file ranges at level-L will usually not be aligned with the file
ranges at level-L+1). The compaction will therefore read 26MB and write 26MB.
Assuming a disk IO rate of 100MB/s (ballpark range for modern drives), the worst
compaction cost will be approximately 0.5 second.

If we throttle the background writing to something small, say 10% of the full
100MB/s speed, a compaction may take up to 5 seconds. If the user is writing at
10MB/s, we might build up lots of level-0 files (~50 to hold the 5*10MB). This
may significantly increase the cost of reads due to the overhead of merging more
files together on every read.

Solution 1: To reduce this problem, we might want to increase the log switching
threshold when the number of level-0 files is large. Though the downside is that
the larger this threshold, the more memory we will need to hold the
corresponding memtable.

Solution 2: We might want to decrease write rate artificially when the number of
level-0 files goes up.

Solution 3: We work on reducing the cost of very wide merges. Perhaps most of
the level-0 files will have their blocks sitting uncompressed in the cache and
we will only need to worry about the O(N) complexity in the merging iterator.

Number of files Copy Section

Instead of always making 2MB files, we could make larger files for larger levels
to reduce the total file count, though at the expense of more bursty
compactions. Alternatively, we could shard the set of files into multiple
directories.

An experiment on an ext3 filesystem on Feb 04, 2011 shows the following timings
to do 100K file opens in directories with varying number of files:

So maybe even the sharding is not necessary on modern filesystems?

Recovery Copy Chapter

• Read CURRENT to find name of the latest committed MANIFEST
• Read the named MANIFEST file
• Clean up stale files
• We could open all sstables here, but it is probably better to be lazy...
• Convert log chunk to a new level-0 sstable
• Start directing new writes to a new log file with recovered sequence#

Garbage collection of files Copy Chapter

RemoveObsoleteFiles() is called at the end of every compaction and at the end
of recovery. It finds the names of all files in the database. It deletes all log
files that are not the current log file. It deletes all table files that are not
referenced from some level and are not the output of an active compaction.

Doc Log Format (`doc/log_format.md`) Copy Section

leveldb Log format

The log file contents are a sequence of 32KB blocks. The only exception is that
the tail of the file may contain a partial block.

Each block consists of a sequence of records:

block := record* trailer?
record :=
  checksum: uint32     // crc32c of type and data[] ; little-endian
  length: uint16       // little-endian
  type: uint8          // One of FULL, FIRST, MIDDLE, LAST
  data: uint8[length]

A record never starts within the last six bytes of a block (since it won't fit).
Any leftover bytes here form the trailer, which must consist entirely of zero
bytes and must be skipped by readers.

Aside: if exactly seven bytes are left in the current block, and a new non-zero
length record is added, the writer must emit a FIRST record (which contains zero
bytes of user data) to fill up the trailing seven bytes of the block and then
emit all of the user data in subsequent blocks.

More types may be added in the future. Some Readers may skip record types they
do not understand, others may report that some data was skipped.

FULL == 1
FIRST == 2
MIDDLE == 3
LAST == 4

The FULL record contains the contents of an entire user record.

FIRST, MIDDLE, LAST are types used for user records that have been split into
multiple fragments (typically because of block boundaries). FIRST is the type
of the first fragment of a user record, LAST is the type of the last fragment of
a user record, and MIDDLE is the type of all interior fragments of a user
record.

Example: consider a sequence of user records:

A: length 1000
B: length 97270
C: length 8000

A will be stored as a FULL record in the first block.

B will be split into three fragments: first fragment occupies the rest of
the first block, second fragment occupies the entirety of the second block, and
the third fragment occupies a prefix of the third block. This will leave six
bytes free in the third block, which will be left empty as the trailer.

C will be stored as a FULL record in the fourth block.

Some benefits over the recordio format: Copy Chapter

1. We do not need any heuristics for resyncing - just go to next block boundary
   and scan. If there is a corruption, skip to the next block. As a
   side-benefit, we do not get confused when part of the contents of one log
   file are embedded as a record inside another log file.

2. Splitting at approximate boundaries (e.g., for mapreduce) is simple: find the
   next block boundary and skip records until we hit a FULL or FIRST record.

3. We do not need extra buffering for large records.

Some downsides compared to recordio format: Copy Chapter

1. No packing of tiny records. This could be fixed by adding a new record type,
   so it is a shortcoming of the current implementation, not necessarily the
   format.

2. No compression. Again, this could be fixed by adding new record types.

Doc Table Format (`doc/table_format.md`) Copy Section

leveldb File format

[data block 1]
[data block 2]
...
[data block N]
[meta block 1]
...
[meta block K]
[metaindex block]
[index block]
[Footer]        (fixed size; starts at file_size - sizeof(Footer))
<end_of_file>

The file contains internal pointers. Each such pointer is called
a BlockHandle and contains the following information:

offset:   varint64
size:     varint64

See varints
for an explanation of varint64 format.

1. The sequence of key/value pairs in the file are stored in sorted
   order and partitioned into a sequence of data blocks. These blocks
   come one after another at the beginning of the file. Each data block
   is formatted according to the code in block_builder.cc, and then
   optionally compressed.

2. After the data blocks we store a bunch of meta blocks. The
   supported meta block types are described below. More meta block types
   may be added in the future. Each meta block is again formatted using
   block_builder.cc and then optionally compressed.

3. A "metaindex" block. It contains one entry for every other meta
   block where the key is the name of the meta block and the value is a
   BlockHandle pointing to that meta block.

4. An "index" block. This block contains one entry per data block,
   where the key is a string >= last key in that data block and before
   the first key in the successive data block. The value is the
   BlockHandle for the data block.

5. At the very end of the file is a fixed length footer that contains
   the BlockHandle of the metaindex and index blocks as well as a magic number.

   text Copy

   metaindex_handle: char[p];     // Block handle for metaindex
    index_handle:     char[q];     // Block handle for index
    padding:          char[40-p-q];// zeroed bytes to make fixed length
                                   // (40==2*BlockHandle::kMaxEncodedLength)
    magic:            fixed64;     // == 0xdb4775248b80fb57 (little-endian)

"filter" Meta Block Copy Chapter

If a FilterPolicy was specified when the database was opened, a
filter block is stored in each table. The "metaindex" block contains
an entry that maps from filter.<N> to the BlockHandle for the filter
block where <N> is the string returned by the filter policy's
Name() method.

The filter block stores a sequence of filters, where filter i contains
the output of FilterPolicy::CreateFilter() on all keys that are stored
in a block whose file offset falls within the range

[ i*base ... (i+1)*base-1 ]

Currently, "base" is 2KB. So for example, if blocks X and Y start in
the range [ 0KB .. 2KB-1 ], all of the keys in X and Y will be
converted to a filter by calling FilterPolicy::CreateFilter(), and the
resulting filter will be stored as the first filter in the filter
block.

The filter block is formatted as follows:

[filter 0]
[filter 1]
[filter 2]
...
[filter N-1]

[offset of filter 0]                  : 4 bytes
[offset of filter 1]                  : 4 bytes
[offset of filter 2]                  : 4 bytes
...
[offset of filter N-1]                : 4 bytes

[offset of beginning of offset array] : 4 bytes
lg(base)                              : 1 byte

The offset array at the end of the filter block allows efficient
mapping from a data block offset to the corresponding filter.

"stats" Meta Block Copy Chapter

This meta block contains a bunch of stats. The key is the name
of the statistic. The value contains the statistic.

TODO(postrelease): record following stats.

data size
index size
key size (uncompressed)
value size (uncompressed)
number of entries
number of data blocks

--- METRICS ---

• Files Extracted: 6
• Estimated Token Budget: ~11155 tokens
• Canonical Reference: https://codewiki.google/github.com/google/leveldb