SAS In-Memory Statistics
for Hadoop is an offering that provides the data scientist or analytical
expert with interactive programming access to in-memory data and integrates
seamlessly with Hadoop.
In order to use the
offering, the following must be true:
-
You are using a distributed SAS LASR Analytic Server
only.
-
The SAS LASR Analytic Server
is co-located with SAS High-Performance Deployment of Hadoop
or a commercial Hadoop distribution that has been configured with
the services from SAS High-Performance Deployment of Hadoop.
The services enable you to use the SASHDAT file format for storing
tables in HDFS.
-
SAS/ACCESS Interface to Hadoop
is configured on a client machine that you use for submitting SAS
programs. Be sure to install the SAS Embedded Process on the machines
in the Hadoop cluster. The
SAS/ACCESS engine, the embedded process,
and the HDMD procedure enable you to describe your data that is in
Hadoop and access it directly without an intermediate metadata repository
such as Hive.
-
SAS Studio provides an interactive
web-based development application that enables you to write and submit
SAS programs. Make sure that your user ID is configured for passwordless
SSH to the machines the cluster. Also make sure that you have passwordless
SSH access from the machine that hosts SAS Studio to the machines
in the cluster.
For more information,
see About Passwordless SSH and SAS Studio.