Options for a Client and Server Running on the Same UNIX Machine

NETCOMP=

Summary

Compresses the data stream for an SPD Server network packet.

Syntax

NETCOMP=YES | NO 
Arguments
YES
sends compressed data in an SPD Server network packet.
NO
sends uncompressed data in an SPD Server network packet.

Description

Normally, data compression for inter-process transfers is recommended. However, for a client and server process on the same machine -- with UNIXDOMAIN=YES -- turning off compression can improve performance. You should examine NETCOMP together with UNIXDOMAIN and NETPACKSIZE for both client and server on the same machine.

Example

Specify to turn off compression of the data stream.
LIBNAME mylib sasspds 'test_area'
   netcomp=no;

UNIXDOMAIN=

Summary

Specifies the use of UNIX domain sockets for data communication between an SPD Server and client process running on the same machine. (Not available in Windows.)

Syntax

UNIXDOMAIN=YES | NO 
Arguments
YES
uses AF_UNIX domain sockets for client/server data communication.
NO
uses the default AF_INET domain sockets for client/server data communication.

Description

When UNIXDOMAIN=YES, SPD Server uses AF_UNIX domain sockets rather than the customary AF_INET domain sockets for data communication. AF_UNIX sockets typically are much faster and greatly enhance performance but are possible only for cases where client and server are running on the same machine. You should also examine NETCOMP and NETPACKSIZE parameters for possible use to enhance performance in conjunction with UNIXDOMAIN.

Example

You find that using the AF_UNIX sockets for your session that is running on the same machine as the SPD Server is not faster. Configure SPD Server to use the default AF_INET sockets instead.
LIBNAME mylib sasspds 'test_area'
   unixdomain=no;
Note: If you are running SPD Server 4.5 or later, and the client and server are both running UNIX, SPD Server automatically detects UNIX domain sockets. In such cases, it is not necessary to specify the UNIXDOMAIN parameter for optimum performance.