Embedded databases benchmarks

benchmarks.md

@coleifer: Did some simple benchmarking of embedded databases in python. It was done in with the following environment:

• Linux 3.14.4
• Python 2.7.7 (Py2K 4 lyfe!)
• SSD

I have modified his script for python 3.5 and run the script against sd card. My environment is:

• Linux 4.4.38
• Python 3.5.2
• SD card
• ARM v7

Testing with N = 100000

UnQLite

Writes: 2.444714069366455 Reads: 1.6616613864898682

UnQLiteJSON

Writes: 3.617347478866577 Reads: 2.1200473308563232

Vedis

Writes: 2.357679605484009 Reads: 1.624964714050293

VedisKV

Writes: 2.47371244430542 Reads: 1.6631860733032227

GDBM

Writes: 3.0294344425201416 Reads: 1.49436616897583

BDBBTree

Writes: 8.086156606674194 Reads: 4.68268609046936

BDBHashTable

Writes: 9.237087488174438 Reads: 5.801295757293701

KyotoBTree

Writes: 1.4112212657928467 Reads: 1.4075307846069336

KyotoHashTable

Writes: 1.5034441947937012 Reads: 1.3784515857696533

LevelDB

Writes: 2.7352421283721924 Reads: 1.335801362991333

Sqlite

Writes: 1267.9827723503113 Reads: 7.064197301864624

run-bench.py

import os
import time

from unqlite import UnQLite
from vedis import Vedis
import bsddb3
import dbm.gnu as gdbm
import kyotocabinet as kc
import plyvel
import sqlite3


def timed(fn):
    def inner(*args):
        s = time.time()
        fn(*args)
        return time.time() - s
    return inner

class Test(object):
    def __init__(self, n):
        self.n = n
        if os.path.exists(self.filename):
            if os.path.isdir(self.filename):
                os.removedirs(self.filename)
            else:
                os.unlink(self.filename)
        self._db = self.get_db()

    def close(self):
        try:
            self._db.close()
        except AttributeError:
            pass

    @timed
    def test_writes(self):
        for i in range(self.n):
            self.write(str(i))

    @timed
    def test_reads(self):
        for i in range(self.n):
            self.read(str(i))

    def write(self, i):
        self._db[i] = i

    def read(self, i):
        return self._db[i]

class UnQLiteTest(Test):
    filename = '/tmp/unq.db'

    def get_db(self):
        return UnQLite(self.filename)

class UnQLiteJSONTest(Test):
    filename = '/tmp/unq-json.db'

    def get_db(self):
        db = UnQLite(self.filename)
        self.coll = db.collection('values')
        self.coll.create()

    @timed
    def test_writes(self):
        data = [{str(i): i} for i in range(self.n)]
        self.coll.store(data)

    @timed
    def test_reads(self):
        return self.coll.all()

class VedisTest(Test):
    filename = '/tmp/vedis.db'

    def get_db(self):
        return Vedis(self.filename)

class VedisKVTest(Test):
    filename = '/tmp/vediskv.db'

    def write(self, i):
        self._db.store(i, i)

    def read(self, i):
        return self._db.fetch(i)

    def get_db(self):
        return Vedis(self.filename)

class GDBMTest(Test):
    filename = '/tmp/gdbm.db'

    def get_db(self):
        return gdbm.open(self.filename, 'c')

class BDBBTreeTest(Test):
    filename = '/tmp/bdb_btree.db'

    def get_db(self):
        return bsddb3.btopen(self.filename)

    def write(self, i):
        self._db[bytes(i, "utf-8")] = bytes(i, "utf-8")

    def read(self, i):
        return self._db[bytes(i, "utf-8")]


class BDBHashTableTest(Test):
    filename = '/tmp/bdb_hashtable.db'

    def get_db(self):
        return bsddb3.hashopen(self.filename)

    def write(self, i):
        self._db[bytes(i, "utf-8")] = bytes(i, "utf-8")

    def read(self, i):
        return self._db[bytes(i, "utf-8")]

class KyotoBTreeTest(Test):
    filename = '/tmp/kyoto_btree.kct'

    def get_db(self):
        db = kc.DB()
        db.open(self.filename, kc.DB.OWRITER | kc.DB.OCREATE)
        return db

    def write(self, i):
        self._db.set(i, i)

    def read(self, i):
        return self._db.get(i)

class KyotoHashTableTest(Test):
    filename = '/tmp/kyoto_hashtable.kch'

    def get_db(self):
        db = kc.DB()
        db.open(self.filename, kc.DB.OWRITER | kc.DB.OCREATE)
        return db

    def write(self, i):
        self._db.set(i, i)

    def read(self, i):
        return self._db.get(i)

class LevelDBTest(Test):
    filename = '/tmp/leveldb.db'

    def get_db(self):
        return plyvel.DB(self.filename, create_if_missing=True)

    def write(self, i):
        self._db.put(bytes(i, "utf-8"), bytes(i, "utf-8"))

    def read(self, i):
        return self._db.get(bytes(i, "utf-8"))

class RocksDBTest(Test):
    filename = '/tmp/rocksdb.db'

    def get_db(self):
        return rocksdb.DB(self.filename,
                          rocksdb.Options(create_if_missing=True))

    def write(self, i):
        self._db.put(i, i)

    def read(self, i):
        self._db.get(i)

class SqliteTest(Test):
    filename = '/tmp/sqlite.db'

    def get_db(self):
        db = sqlite3.connect(self.filename)
        self._cursor = db.cursor()
        self._cursor.execute("""
            create table data (
                key varchar(255) not null primary key,
                value varchar(255) not null
            );
        """)
        return db

    def write(self, i):
        self._cursor.execute("INSERT INTO data (key, value) VALUES (?, ?)",
                             (i, i))
        self._db.commit()

    def read(self, i):
        self._cursor.execute("SELECT * FROM data WHERE key=?", (i,))
        self._cursor.fetchall()

class RedisTest(Test):
    filename = '/tmp/na'

    def get_db(self):
        db = Redis(db=15)
        db.flushdb()
        return db

tests = [
    UnQLiteTest,
    UnQLiteJSONTest,
    VedisTest,
    VedisKVTest,
    GDBMTest,
    BDBBTreeTest,
    BDBHashTableTest,
    KyotoBTreeTest,
    KyotoHashTableTest,
    LevelDBTest,
    SqliteTest,
]

N = 100000

print('Testing with N = %s' % N )
print( '------------------------------------' )
print('\n\n')

for TestClass in tests:
    test_name = TestClass.__name__.replace('Test', '')
    print( test_name )
    print( '~' * len(test_name))
    test = TestClass(N)
    print( 'Writes: ', test.test_writes())
    print( 'Reads:  ', test.test_reads())
    print('\n')
    test.close()