Installation

Quickstart

Below are instructions to quickly set up an environment to test Omid in your local machine.

Requirements

1. Java 7
2. HBase 0.98

TSO Setup

1. Download and Install the Required HBase Version

You can find HBase distributions in this page. Then start HBase in standalone mode.

2. Clone the Omid repository and Build the TSO Package:

$ git clone [email protected]:yahoo/omid.git
$ cd omid
$ mvn clean install assembly:single

This will generate a binary package containing all dependencies for the TSO in tso-server/target/tso-server-<VERSION>-bin.tar.gz. You can use -DskipTests in the maven command to avoid passing the test suite. Unit tests coverage is quite extensive and take a while to run on each build (~40min at the moment of writing). So, consider using mvn clean install -DskipTests to speed temporal builds. Note that -Dmaven.test.skip=true is NOT an equivalent.

As an alternative to clone the project, you can download the required version for the TSO tar.gz package from the release repository.

You can also see the build history here.

3. Extract the TSO Package

$ tar zxvf tso-server-<VERSION>-bin.tar.gz
$ cd tso-server-<VERSION>

4. Create Omid Tables

Ensure that the setting for hbase.zookeeper.quorum in conf/hbase-site.xml points to your zookeeper instance, and create the Timestamp Table and the Commit Table using the omid.sh script included in the bin directory of the tso server:

$ bin/omid.sh create-hbase-commit-table -numSplits 16
$ bin/omid.sh create-hbase-timestamp-table

These two tables are required by Omid and they must not be accessed by client applications.

5. Start the TSO Server

$ bin/omid.sh tso

This starts the TSO server that in turn will connect to HBase to store information in HBase. By default the TSO listens on port 54758. If you want to change the TSO configuration, you can modify the contents in the conf/omid.conf file

HBase Client Usage

1. Create a New Java Application

Use your favorite IDE an create a new project.

2. Add hbase-client Dependency

Choose the right version of the hbase-client jar. For example, in a Maven-based app add the following dependency in the pom.xml file:

<dependency>
   <groupId>com.yahoo.omid</groupId>
   <artifactId>hbase-client</artifactId>
   <version>${hbase_client.version}</version>
</dependency>

3. Start Coding Using the Omid Client Interfaces

In Omid there are two client interfaces: TTable and TransactionManager (These interfaces will likely change slightly in future.):

1. The TransactionManager is used for creating transactional contexts, that is, transactions. A builder is provided in the HBaseTransactionManager class in order to get the TransactionManager interface.

2. TTable is used for putting, getting and scanning entries in a HBase table. TTable's interface is similar to the standard HTableInterface, and only requires passing the transactional context as a first parameter in the transactional aware methods (e.g. put(Transaction tx, Put put))

Example Application

Below is provided a sample application accessing data transactionally. Its a dummy application that writes two cells in two different rows of a table in a transactional context, but is enough to show how the different Omid client APIs are used. A detailed explanation of the client interfaces can be found in the Basic Examples section.

import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.hbase.HBaseConfiguration;
import org.apache.hadoop.hbase.client.Put;
import org.apache.hadoop.hbase.util.Bytes;
import com.yahoo.omid.transaction.HBaseTransactionManager;
import com.yahoo.omid.transaction.TTable;
import com.yahoo.omid.transaction.Transaction;
import com.yahoo.omid.transaction.TransactionManager;


public class OmidExample {
    public static void main(String[] args) throws Exception {
        Configuration conf = HBaseConfiguration.create();
        conf.set("tso.host", "localhost");
        conf.setInt("tso.port", 54758);

        TransactionManager tm = HBaseTransactionManager.newBuilder()
                                                       .withConfiguration(conf)
                                                       .build();

        TTable tt = new TTable(conf, "MY_TX_TABLE");
        byte[] exampleRow1 = Bytes.toBytes("EXAMPLE_ROW1");
        byte[] exampleRow2 = Bytes.toBytes("EXAMPLE_ROW2");
        byte[] family = Bytes.toBytes("MY_CF");
        byte[] qualifier = Bytes.toBytes("foo");
        byte[] dataValue1 = Bytes.toBytes("val1");
        byte[] dataValue2 = Bytes.toBytes("val2");

        Transaction tx = tm.begin();
        Put row1 = new Put(exampleRow1);
        row1.add(family, qualifier, dataValue1);
        tt.put(tx, row1);
        Put row2 = new Put(exampleRow2);
        row2.add(family, qualifier, dataValue2);
        tt.put(tx, row2);
        tm.commit(tx);

        tt.close();
        tm.close();
    }
}

To run the application, make sure core-site.xml and hbase-site.xml for your HBase cluster are present in your CLASSPATH. You will need to set tso.host and tso.port appropriately. Also, you will need to create a HBase table "MY_TX_TABLE", with column family "MY_CF", and with TTL disabled and VERSIONS set to Integer.MAX_VALUE. For example using the HBase shell:

create 'MY_TX_TABLE', {NAME => 'MY_CF', VERSIONS => '2147483647', TTL => '2147483647'}

This example assumes non-secure communication with HBase. If your HBase cluster is secured with Kerberos, you will need to use the UserGroupInformation API to log in securely.

The Omid Compactor Coprocessor

Omid includes a jar with an HBase coprocessor for performing data cleanup that operates during compactions, both minor and major. Specifically, it does the following:

• Cleans up garbage data from aborted transactions
• Purges deleted cells. Omid deletes work by placing a special tombstone marker in cells. The compactor detects these and actually purges data when it is safe to do so (i.e. when there are no committable transactions that may read the data).
• 'Heals' committed cells for which the writer failed to write shadow cells.

To deploy the coprocessor, the coprocessor jar must be placed in a location (typically on HDFS) that is accessible by HBase region servers. The coprocessor may then be enabled on a transactional table by the following steps in the HBase shell:

1) Disable the table

disable 'MY_TX_TABLE'

2) Add a coprocessor specification to the table via a "coprocessor" attribute. The coprocessor spec may (and usually will) also include the name of the Omid commit table

alter 'MY_TX_TABLE', METHOD => 'table_att', 'coprocessor'=>'<path_to_omid_coprocessor>/omid-hbase-coprocessor-<coprocessor_version>.jar|com.yahoo.omid.transaction.OmidCompactor|1001|omid.committable.tablename=OMID_COMMIT_TABLE'

3) Add an "OMID_ENABLED => true" flag to any column families which the co-processor should work on

alter 'MY_TX_TABLE', { NAME => 'MY_CF', METADATA =>  {'OMID_ENABLED' => 'true'}}

4) Re-enable the table

enable 'MY_TX_TABLE'