Setting Up a New Cluster

The configuration and deployment of a new cluster requires these basic steps:

Setting Up an External Presto Configuration Folder: this allows you to keep most of the configuration and extensions for Presto in a single set of folders that can be shared across the entire cluster. This simplifies both the initial configuration as well as ongoing updates and deployment of new mashables, mashups or apps.

Note:

This step is highly recommended, but not required. If you do not use a shared configuration folder, all subsequent updates to configuration or extensions for new artifacts must be manually copied to each member of the cluster.

This folder should reside in a file system that is shared or mounted across the cluster. You may also need to provide data redundancy or failover capabilities for this shared file system.

As part of this step, you also typically deploy one Presto Server in the cluster and complete most of the basic configuration that will be shared across the cluster.

Sharing the Presto Repository in Clustered Environments: all nodes in the cluster work with a shared Presto Repository which you must create and configure.

Sharing the Presto Repository does not, by itself, provide any data redundancy, load balancing or failover capabilities for the database. These requirements are handled in the data layer by your database server or other replication/synchronization solutions, such as DRBD. For more information, see documentation for your database or replication/synchronization solution.

Configuring Caching for the Cluster: each Presto Server has a local cache for mashable and mashup responses as well as local caches for updates to artifacts. If Presto Analytics is enabled in your Presto license, the Presto Analytics In-Memory Stores are also local.

Can leave the response cache as a local cache or you can configure a distributed cache that all Presto Servers in the cluster share.

Must configure a distributed cache for artifact updates that all Presto Servers in the cluster share.

Must configure a distributed cache for the Presto Analytics In-Memory Stores that all Presto Servers in the cluster share.

See Configure BigMemory Servers for Presto Caching and In-Memory Stores for instructions on how to configure BigMemory, or other caching solutions, as a distributed cache for .Presto

Defining the Application Server Cluster: the application servers that host each Presto Server define and handle clustering requirements at the application layer. You can also add a load balancer to the cluster.

In addition to the basic cluster configuration required by your application server and load balancer, Presto has a single requirement for application-layer cluster configuration. You must either:

Enable session affinity, sometimes also called ’sticky sessions,’ in the load balancer.

See documentation for your application server and/or load balancer for information on how to do this.

Adding Additional Presto Servers to the Cluster: once you have set up the shared resources, you can deploy and add additional members to the cluster. See Adding New Members to an Existing Cluster for instructions.

Add MetaData and Deploy Artifacts: for this new environment. For artifacts, you can automate some parts of this process using export and import commands. See Deploying Presto Artifacts and Other Metadata for instructions.