ceph/src/rocksdb/docs/_posts/2019-08-15-unordered-write.markdown

   1 ---
   2 title: Higher write throughput with `unordered_write` feature
   3 layout: post
   4 author: maysamyabandeh
   5 category: blog
   6 ---
   7
   8 Since RocksDB 6.3, The `unordered_write=`true option together with WritePrepared transactions offers 34-42% higher write throughput compared to vanilla RocksDB. If the application can handle more relaxed ordering guarantees, the gain in throughput would increase to 63-131%.
   9
  10 ### Background
  11
  12 Currently RocksDB API delivers the following powerful guarantees:
  13 - Atomic reads: Either all of a write batch is visible to reads or none of it.
  14 - Read-your-own writes: When a write thread returns to the user, a subsequent read by the same thread will be able to see its own writes.
  15 - Immutable Snapshots: The reads visible to the snapshot are immutable in the sense that it will not be affected by any in-flight or future writes.
  16
  17 ### `unordered_write`
  18
  19 The `unordered_write` feature, when turned on, relaxes the default guarantees of RocksDB. While it still gives read-your-own-write property, neither atomic reads nor the immutable snapshot properties are provided any longer. However, RocksDB users could still get read-your-own-write and immutable snapshots when using this feature in conjunction with TransactionDB configured with WritePrepared transactions and `two_write_queues`. You can read [here](https://github.com/facebook/rocksdb/wiki/unordered_write) to learn about the design of `unordered_write` and [here](https://github.com/facebook/rocksdb/wiki/WritePrepared-Transactions) to learn more about WritePrepared transactions.
  20
  21 ### How to use it?
  22
  23 To get the same guarantees as vanilla RocksdB:
  24
  25     DBOptions db_options;
  26     db_options.unordered_write = true;
  27     db_options.two_write_queues = true;
  28     DB* db;
  29     {
  30       TransactionDBOptions txn_db_options;
  31       txn_db_options.write_policy = TxnDBWritePolicy::WRITE_PREPARED;
  32       txn_db_options.skip_concurrency_control = true;
  33       TransactionDB* txn_db;
  34       TransactionDB::Open(options, txn_db_options, kDBPath, &txn_db);
  35       db = txn_db;
  36     }
  37     db->Write(...);
  38
  39 To get relaxed guarantees:
  40
  41     DBOptions db_options;
  42     db_options.unordered_write = true;
  43     DB* db;
  44     DB::Open(db_options, kDBPath, &db);
  45     db->Write(...);
  46
  47 # Benchmarks
  48
  49     TEST_TMPDIR=/dev/shm/ ~/db_bench --benchmarks=fillrandom --threads=32 --num=10000000 -max_write_buffer_number=16 --max_background_jobs=64 --batch_size=8 --writes=3000000 -level0_file_num_compaction_trigger=99999 --level0_slowdown_writes_trigger=99999 --level0_stop_writes_trigger=99999 -enable_pipelined_write=false -disable_auto_compactions --transaction_db=true --unordered_write=1 --disable_wal=0
  50
  51 Throughput with `unordered_write`=true and using WritePrepared transaction:
  52 - WAL: +42%
  53 - No-WAL: +34%
  54 Throughput with `unordered_write`=true
  55 - WAL: +63%
  56 - NoWAL: +131%