YCSB/jdbc
@FranckPachot 25724e8f2c [jdbc] Add support for SQL:2008 FETCH FIRST n ROWS ONLY for Scans (#1373) 2019-11-13 13:50:44 -06:00
..
src [jdbc] Add support for SQL:2008 FETCH FIRST n ROWS ONLY for Scans (#1373) 2019-11-13 13:50:44 -06:00
README.md Move all packages from com.yahoo.ycsb to site.ycsb. (#1354) 2019-09-19 13:51:09 -05:00
pom.xml [version] update master branch to 0.18.0-SNAPSHOT 2019-09-21 01:24:50 -05:00

README.md

JDBC Driver for YCSB

This driver enables YCSB to work with databases accessible via the JDBC protocol.

Getting Started

1. Start your database

This driver will connect to databases that use the JDBC protocol, please refer to your databases documentation on information on how to install, configure and start your system.

2. Set up YCSB

You can clone the YCSB project and compile it to stay up to date with the latest changes. Or you can just download the latest release and unpack it. Either way, instructions for doing so can be found here: https://github.com/brianfrankcooper/YCSB.

3. Configure your database and table.

You can name your database what ever you want, you will need to provide the database name in the JDBC connection string.

You can name your table whatever you like also, but it needs to be specified using the YCSB core properties, the default is to just use 'usertable' as the table name.

The expected table schema will look similar to the following, syntactical differences may exist with your specific database:

CREATE TABLE usertable (
	YCSB_KEY VARCHAR(255) PRIMARY KEY,
	FIELD0 TEXT, FIELD1 TEXT,
	FIELD2 TEXT, FIELD3 TEXT,
	FIELD4 TEXT, FIELD5 TEXT,
	FIELD6 TEXT, FIELD7 TEXT,
	FIELD8 TEXT, FIELD9 TEXT
);

Key take aways:

  • The primary key field needs to be named YCSB_KEY
  • The other fields need to be prefixed with FIELD and count up starting from 1
  • Add the same number of FIELDs as you specify in the YCSB core properties, default is 10.
  • The type of the fields is not so important as long as they can accept strings of the length that you specify in the YCSB core properties, default is 100.

JdbcDBCreateTable Utility

YCSB has a utility to help create your SQL table. NOTE: It does not support all databases flavors, if it does not work for you, you will have to create your table manually with the schema given above. An example usage of the utility:

java -cp YCSB_HOME/jdbc-binding/lib/jdbc-binding-0.4.0.jar:mysql-connector-java-5.1.37-bin.jar site.ycsb.db.JdbcDBCreateTable -P db.properties -n usertable

Hint: you need to include your Driver jar in the classpath as well as specify JDBC connection information via a properties file, and a table name with -n.

Simply executing the JdbcDBCreateTable class without any other parameters will print out usage information.

4. Configure YCSB connection properties

You need to set the following connection configurations:

db.driver=com.mysql.jdbc.Driver
db.url=jdbc:mysql://127.0.0.1:3306/ycsb
db.user=admin
db.passwd=admin

Be sure to use your driver class, a valid JDBC connection string, and credentials to your database.

For connection fail-over in a DBMS cluster specify the connection string as follows (example based on Postgres):

db.driver=org.postgresql.Driver
db.url=jdbc:postgresql://IP1:PORT1,IP2:PORT2,IP3:PORT3/ycsb
db.user=admin
db.passwd=admin

For using multiple shards in a DBMS cluster specify the connection string as follows by using ;as delimiter (example based on PostgreSQL):

db.driver=org.postgresql.Driver
db.url=jdbc:postgresql://host1:port1/ycsb;jdbc:postgresql://host2:port2/ycsb
db.user=admin
db.passwd=admin

You can add these to your workload configuration or a separate properties file and specify it with -P or you can add the properties individually to your ycsb command with -p.

5. Add your JDBC Driver to the classpath

There are several ways to do this, but a couple easy methods are to put a copy of your Driver jar in YCSB_HOME/jdbc-binding/lib/ or just specify the path to your Driver jar with -cp in your ycsb command.

6. Running a workload

Before you can actually run the workload, you need to "load" the data first.

bin/ycsb load jdbc -P workloads/workloada -P db.properties -cp mysql-connector-java.jar

Then, you can run the workload:

bin/ycsb run jdbc -P workloads/workloada -P db.properties -cp mysql-connector-java.jar

Configuration Properties

db.driver=com.mysql.jdbc.Driver				# The JDBC driver class to use.
db.url=jdbc:mysql://127.0.0.1:3306/ycsb		# The Database connection URL.
db.user=admin								# User name for the connection.
db.passwd=admin								# Password for the connection.
db.batchsize=1000             # The batch size for doing batched inserts. Defaults to 0. Set to >0 to use batching.
jdbc.fetchsize=10							# The JDBC fetch size hinted to the driver.
jdbc.autocommit=true						# The JDBC connection auto-commit property for the driver.
jdbc.batchupdateapi=false     # Use addBatch()/executeBatch() JDBC methods instead of executeUpdate() for writes (default: false)
db.batchsize=1000             # The number of rows to be batched before commit (or executeBatch() when jdbc.batchupdateapi=true)

Please refer to https://github.com/brianfrankcooper/YCSB/wiki/Core-Properties for all other YCSB core properties.

JDBC Parameter to Improve Insert Performance

Some JDBC drivers support re-writing batched insert statements into multi-row insert statements. This technique can yield order of magnitude improvement in insert statement performance. To enable this feature:

  • db.batchsize must be greater than 0. The magniute of the improvement can be adjusted by varying batchsize. Start with a small number and increase at small increments until diminishing return in the improvement is observed.
  • set jdbc.batchupdateapi=true to enable batching.
  • set JDBC driver specific connection parameter in db.url to enable the rewrite as shown in the examples below: