mirror of
https://github.com/facebook/rocksdb.git
synced 2024-11-27 02:44:18 +00:00
4e83b8001a
Summary: Closes https://github.com/facebook/rocksdb/pull/2219 Differential Revision: D4986642 Pulled By: lightmark fbshipit-source-id: c9328991e742768fb5caa0e88e022afb514f0c65
2.3 KiB
2.3 KiB
title | layout | author | category |
---|---|---|---|
Bulkloading by ingesting external SST files | post | IslamAbdelRahman | blog |
Introduction
One of the basic operations of RocksDB is writing to RocksDB, Writes happen when user call (DB::Put, DB::Write, DB::Delete ... ), but what happens when you write to RocksDB ? .. this is a brief description of what happens.
- User insert a new key/value by calling DB::Put() (or DB::Write())
- We create a new entry for the new key/value in our in-memory structure (memtable / SkipList by default) and we assign it a new sequence number.
- When the memtable exceeds a specific size (64 MB for example), we convert this memtable to a SST file, and put this file in level 0 of our LSM-Tree
- Later, compaction will kick in and move data from level 0 to level 1, and then from level 1 to level 2 .. and so on
But what if we can skip these steps and add data to the lowest possible level directly ? This is what bulk-loading does
Bulkloading
- Write all of our keys and values into SST file outside of the DB
- Add the SST file into the LSM directly
This is bulk-loading, and in specific use-cases it allow users to achieve faster data loading and better write-amplification.
and doing it is as simple as
Options options;
SstFileWriter sst_file_writer(EnvOptions(), options, options.comparator);
Status s = sst_file_writer.Open(file_path);
assert(s.ok());
// Insert rows into the SST file, note that inserted keys must be
// strictly increasing (based on options.comparator)
for (...) {
s = sst_file_writer.Add(key, value);
assert(s.ok());
}
// Ingest the external SST file into the DB
s = db_->IngestExternalFile({"/home/usr/file1.sst"}, IngestExternalFileOptions());
assert(s.ok());
You can find more details about how to generate SST files and ingesting them into RocksDB in this wiki page
Use cases
There are multiple use cases where bulkloading could be useful, for example
- Generating SST files in offline jobs in Hadoop, then downloading and ingesting the SST files into RocksDB
- Migrating shards between machines by dumping key-range in SST File and loading the file in a different machine
- Migrating from a different storage (InnoDB to RocksDB migration in MyRocks)