Build & Push Jobs for Voldemort Read-Only
We have been using the Build and Push Job at Linkedin to create Voldemort Read-Only stores from data present in sequence files/ Avro Container files on HDFS. The Voldemort build and push job uses the fault tolerance and parallelism of Hadoop and builds individual Voldemort node/partition-level data stores, which are transferred to Voldemort for serving. A Hadoop job reads data from a source in HDFS, repartitions it on a per-node basis, and ﬁnally writes the data to individual Read-only storage engine .
The VoldemortBuildAndPushJob will behave in the following way:
- Build an XML storeDef for your data (based off of the key and value metadata in your JsonSequenceFile/Avro on HDFS).
- Connect to push.cluster
- Get the storeDefs for all stores in the push.cluster
- Look through the storeDefs for a store with the same name as push.store.name. If one is found, validate that the storeDef in the cluster matches the storeDef for your data. If it doesn’t, fail. If no storeDef exists on the cluster that matches push.store.name, then add your storeDef to the cluster.
- Build the Voldemort store in Hadoop
- Push the Voldemort store to push.cluster
First, you should clone and build the latest version of Voldemort, as described in the Quickstart guide. The "server.properties" file on the Voldemort server needs to have the following entry:
This instructs the server to use that class while fetching files from Hadoop during the push phase.
Then, from the root of the Voldemort git repository, you can generate a fat jar containing all of the dependencies that the Build and Push job needs with the following command:
After that, you can launch the BnP job from your local machine using the following command:
The config_file argument should point to the path of a text file containing configuration parameters for the BnP job. Below are two such example configurations.
Pushing JSON Data - Job File
Pushing AVRO Data - Job File
Notice the following properties:
- build.type.avro=true This specifies that input data is Avro.
- avro.key.field=memberId This specifies the field to be used as the key
- avro.value.field=localizedFirstNames This specifies the field to be used as the value
We use a custom data and index format for the Read-Only store.
On every Node you will find a node directory containing one or mutliple data and index files with the following naming convention: