Configuring for Access to a Data Store with a SAS Embedded Process

The process involved for configuring the SAS High-Performance Analytics environment with a SAS Embedded Process consists of the following steps:

You configure the SAS High-Performance Analytics environment with a SAS Embedded Process using a shell script. The script enables you to configure the environment for the various third-party data stores supported by the SAS Embedded Process.

The analytics environment is designed on the principle, install once, configure many. For example, suppose that your site has three remote data stores from three different third-party vendors whose data you want to analyze. You run the analytics environment configuration script one time and provide the information for each data store vendor as you are prompted for it. (When prompted for a data store vendor that you do not have, simply ignore that set of prompts.)

When you have different versions of the same vendor’s data store, specifying the vendor’s latest client data libraries usually works. However, this choice can be problematic for different versions of Hadoop, where a later set of JAR files is not typically backwardly compatible with earlier versions, or for sites that use Hadoop implementations from more than one vendor. (The configuration script does not delineate between different Hadoop vendors.) In these situations, you must run the analytics environment configuration script once for each different Hadoop version or vendor. As the configuration script creates a TKGrid_REP directory underneath the current directory, it is important to run the script a second time from a different directory.

To illustrate how you might manage configuring the analytics environment for two different Hadoop vendors, consider this example: suppose your site uses Cloudera Hadoop 4 and Hortonworks Data Platform 2. When running the analytics environment script to configure for Cloudera 4, you would create a directory similar to:

cdh4

When configuring the analytics environment for Cloudera, you would run the script from the cdh4 directory. When complete, the script creates a TKGrid_REP child directory:

cdh4/TKGrid_REP

For Hortonworks, you would create a directory similar to:

hdp2

When configuring the analytics environment for Hortonworks, you would run the script from the hdp2 directory. When complete, the script creates a TKGrid_REP child directory:

hdp2/TKGrid_REP