Glossary

Apache Hadoop

a framework that allows for the distributed processing of large data sets across clusters of computers using a simple programming model.

BY-group processing

the process of using the BY statement to process observations that are ordered, grouped, or indexed according to the values of one or more variables. Many SAS procedures and the DATA step support BY-group processing. For example, you can use BY-group processing with the PRINT procedure to print separate reports for different groups of observations in a single SAS data set.

co-located data provider

a distributed data source, such as SAS Visual Analytics Hadoop or a third-party vendor database, that has SAS High-Performance Analytics software installed on the same machines. The SAS software on each machine processes the data that is local to the machine or that the data source makes available as the result of a query.

grid host

the machine to which the SAS client makes an initial connection in a SAS High-Performance Analytics application.

Hadoop Distributed File System

a framework for managing files as blocks of equal size, which are replicated across the machines in a Hadoop cluster to provide fault tolerance.

HDFS

See Hadoop Distributed File System

Message Passing Interface

is a message-passing library interface specification. SAS High-Performance Analytics applications implement MPI for use in high-performance computing environments.

MPI

See Message Passing Interface

root node

in a SAS High-Performance Analytics application, the role of the software that distributes and coordinates the workload of the worker nodes. In most deployments the root node runs on the machine that is identified as the grid host. SAS High-Performance Analytics applications assign the highest MPI rank to the root node.

SASHDAT file

the data format used for tables that are added to HDFS by SAS. SASHDAT files are read in parallel by the server.

server description file

a file that is created by a SAS client when the LASR procedure executes to create a server. The file contains information about the machines that are used by the server. It also contains the name of the server signature file that controls access to the server.

signature file

small files that are created by the server to control access to the server and to the tables loaded in the server. There is one server signature file for each server instance. There is one table signature file for each table that is loaded into memory on a server instance.

worker node

in a SAS High-Performance Analytics application, the role of the software that receives the workload from the root node.