BigMemory for Caching, Connections and Presto Analytics
By default Presto uses local memory for caching and the Presto Analytics In-Memory Stores that Presto Analytics creates. This uses BigMemory as a local client that is installed with Presto and requires only your Presto license. In specific cases, you must also install BigMemory Servers on one or more additional hosts and configure Presto and the Integrated MashZone Server to work with them. Presto requires BigMemory Servers with a BigMemory license to support:
Significant, extensible amounts of memory, most commonly for very large datasets used with
Presto Analytics.
BigMemory Servers can be deployed in clusters, also known as Terracotta Server Arrays, that can easily be extended for scalable memory requirements.
Access to
off-heap memory.
BigMemory Servers also can manage memory outside of heap both for better scalability and performance improvements.
Access to
In-Memory Stores created and populated dynamically by external systems.
BigMemory manages the In-Memory Stores created dynamically by other systems and makes connection information available to Presto through the Terracotta Management Console (TMC) to allow Presto to work with this data. Apama, for example, dynamically creates distributed stores for the Apama MemoryStore which Presto can connect to and query.
Distributed caching when
Presto is deployed in clusters.
With clusters, some of the interal Presto caches must be distributed and managed by BigMemory Servers.
MashZone feeds that use
BigMemory connections as a data source.
See
Configure
BigMemory
Servers for
Presto
Caching and
In-Memory Stores for instructions on configuring
Presto to work with
BigMemory Servers.
See
Working with
Presto Analytics In-Memory Stores for an overview and links to additional information on
Presto Analytics In-Memory Stores.
You also need to provide configuration and connection information for In-Memory Stores that are created by other systems or stores created by Presto Analytics that need different settings than the defaults. For more information on these tasks, see:
