Jump to: navigation, search

Optimizing Performance

Review the recommendations provided in this topic to optimize Stat Server performance.

Hardware-Related Recommendations

For Stat Server operating in regular mode:

  • Consider the following formula, which approximates Stat Server memory, in megabytes, for a typical large contact center:
    MemoryReqd = 100 + (NStatistics × 0.0012)
    where NStatistics represents the number of opened statistics and 0.0012 refers to approximately 1.2 KB of memory per statistic. This formula applies to Stat Server memory calculation of core statistics. Java Extension clients might request additional memory of which Stat Server is unaware.
    For example, Stat Server on a computer with 1.5 GB of memory should be more than ample to handle CC Analyzer requests of 30,000 active Agent or Place objects that originate from the Genesys-provided Agent and Place reports):
    NStatistics = 28 statistics/report layout × 30,000 objects = 840,000 statistics
    MemoryReqd = 100 + (840,000 × 0.0012) = 1,108 MB
    For smaller contact centers, you can reduce the constant (100) to a smaller value.
    Install Stat Server on a computer with sufficient physical memory to avoid swapping.
  • Consider distributing the total number of required statistics for Solution Reporting and real-time interaction processing for all solutions over a number of Stat Server applications.
  • Install Stat Server and source event server on the same computer or connect them through a fast LAN. If you are using several T-Server applications, position Stat Server nearest to the one.
  • Do not install Stat Server on the same computer as Configuration Server.
  • Do not install real-time, third-party applications on the computer that is running Stat Server.
  • If you want to store Stat Server data, consider dedicating a separate Stat Server application whose sole purpose is to write data to the Stat Server database.
  • For large contact centers, consider allocating approximately 5 MB of space for each day Stat Server writes data to a database. This recommendation applies only if you configure Stat Server with a database access point and enable your Stat Server application to write data to a database by setting corresponding configuration options.

For Stat Server operating in cluster mode:

  • Install Stat Server and source event server on the same computer or connect them through a fast LAN. If you are using several SIP Server applications, position Stat Server nearest to the one.
  • The hardware should be powerful enough to handle the entire SIP cluster and Stat Server cluster. The computer should have about at least 32GB of memory and 32 cores.
  • Suggestion: Dedicate one node per processor.
  • For better performance, in addition to the T-Controller connection, you consider configuring more than one Stat Server application each connecting to the same Interaction Proxy in order to better distribute call loads amongst all Stat Server nodes.

Software-Related Recommendations

For Stat Server operating in regular mode:

  • For Stat Server applications that write to the Stat Server database, configure options only for the tables that you need by setting the following configuration options:
    • login-table
    • qinfo-table
    • status-table
    • voice-reasons-table
  • Review the configuration options that are related to write operations to this database:
    • For Oracle, Microsoft SQL, and DB2 relational database management systems (RDBMSs), set the enable-binding option to Yes.
    • Set the local-time-in-status-table configuration option to No if you do not need a translation of UTC time to the time one of the host on which Stat Server is deployed.
    • Set ixn-id-in-status-table to No for Solution Reporting and other clients that employ only an interaction’s connection ID.
  • You can improve Stat Server performance further by tuning the debug-level configuration option. Specify only the debugging log level that you need.

For Stat Server operating in cluster mode:

Stat Server operating in cluster mode supports:

  • 60,000 administered agents (aka regular DNs)
  • 200,000 configured agents
  • 40,000 simultaneously logged-in agents
  • 20,000 skill-based virtual agents groups
  • 10,000 virtual queues
  • 20,000 route points
  • 2,000 agents per agent group
  • 40 calls per second
  • 4.46 million open statistics
  • 60,000 simultaneous calls

One Stat Server instance sustains the load of one T-Controller up to 150 calls per second.

The number of Stat Server instances within a solution need not correlate with the number of SIP Server nodes in a SIP cluster. Instead, this number depends on the following:

  • Call volume
  • Number of Stat Server clients
  • Number of opened statistics

and other factors. Refer to SIP Cluster documentation for other information pertinent to Stat Server performance within the cluster.

You can operate multiple Stat Server solutions on the same host with the restriction that each Stat Server cluster must designate a different Stat Server solution.

This page was last modified on July 4, 2017, at 07:19.


Comment on this article:

blog comments powered by Disqus