Operations Guide

10.0 HotFix 1
- 10.5 HotFix 2
- 10.5 HotFix 1
- 10.5
- 10.2 HotFix 1
- 10.2
- 10.1 HotFix 1
- 10.1
- 10.0

Back Next

Configuration

The following section discusses how to decide which servers are required and how to configure them. Server configuration and placement is important as it affects performance.

The Search and Rulebase Servers are packaged together in one executable image called

ssasrsv

. Based on start up parameters,

ssasrsv

can function as one of the following servers:

Search Server

Rulebase Server

Search and Rulebase Servers

All clients communicate with the servers using a socket interface over TCP/IP. Communication costs vary based on where the client and server are placed. In order of most expensive to least expensive, they are:

remote machines

same machine

same executable process

Clients on remote machines incur the cost of network transmission. Clients on the same machine will take a shortcut through the TCP/IP protocol stack and avoid the transmission costs. A combined Search and Rulebase server takes this one step further by bypassing TCP/IP altogether. Therefore it is advantageous to run a combined Search/Rulebase Server. If more Search Servers are required, they can be started as standalone Search Servers configured to access the Rulebase in the combined server.

Before launching the servers you must decide:

which servers are required

how many Search Servers are required

which machines the servers will run on

The Console Server is required if you want to use the Console Client to administer Identity Resolution. A Connection Server is recommended for stateless search clients. At least, one search Server is required. In general, one Rulebase Server must be started.

Session Pooling Parameters

You can configure the session pooling method by setting the following environment variables in the

env\isss.bat

(Windows) or

env/isss

(UNIX) file:

SSA_SESSION_POOL_MAX: The maximum number of search requests that the Search Server can retain in the pool of search requests. The configuration does not impact the maximum number of search requests that the Search Server can run simultaneously. After the Search Server completes the search requests, it retains the maximum number of search requests that you had set in the pool.; By default, the Search Server retains all the search requests in the pool. You can disable the session pooling method by setting the SSA_SESSION_POOL_MAX variable to 0. Disabling the session pooling method results in the creation and destruction of the search-related resources based on the demand.

SSA_SESSION_POOL_TIMEOUT: The time period for a search request to remain unused in the pool of search requests. After the specified time period, the Search Server releases the unused searches from the pool, which frees the server resources and closes the database connections. You can specify the time period in seconds (s), minutes (m), or hours (h).; The default value is 600 seconds (10 minutes), and the default unit is seconds. For example, if you set the value as SSA_SESSION_POOL_TIMEOUT=3600 or SSA_SESSION_POOL_TIMEOUT=1h, the Search Server releases the unused search requests from the pool after one hour.

SSA_SESSION_POOL_HEARTBEAT: The frequency to run the periodic task that releases the unused search requests. You can specify the time period in seconds (s), minutes (m), or hours (h). The default value is 60 seconds, and the default unit is seconds.

Search Client Limit

The Search Server supports approximately 250 concurrent search clients. The limit is dependent on many factors, including available memory, free sockets, and the type of searches defined in your System. Multi-searches require more internal resources than individual searches, and searches that use SSA-NAME3 require more resources than searches using database indexes (such as Lite-Indexes). In the following discussion, the quoted limit of 250 must be understood as an approximate limit. Your implementation might support many more users, or in some particularly resource intensive cases, somewhat less than 250.

Will there be more than 250 concurrent search clients? If not, you must start a combined Search/Rulebase Server. If you expect more than 250 clients, start a combined Search/Rulebase Server, and one or more standalone Search Servers. It is assumed that your Application Server performs load balancing by distributing search client connection requests between all Search Servers.

Is your database in a cluster? If so, consider running multiple Search Servers and ZooKeeper Servers.

Rename Saved Search

Table of Contents

Operations Guide

Operations Guide

Configuration

Configuration

Session Pooling Parameters

Search Client Limit