Feedback

  • Contents
 

Sizing Calculation

To estimate the appropriate number of servers needed, use the assumptions in Table 1 to establish a comparable set of values.As described in Session Manager Sizing, different types of client applications require different resources.As such, take individual variations from these example usages into account when determining an estimated sizing.

Table 1 User Categories for Sizing

User Category

Description

Business Users

Interaction Desktops loading Company Directory and one additional directory that is a subset of the Company Directory; moderate call volume (less than 20% of users on phone at one time).

Call Center Agents

Interaction Desktops loading Company Directory, one additional directory, Workgroup Queue Statistics; less than 20% of users monitoring one ACD queue; heavy call volume (approximately 100% users on phone or in follow-up; approximately one call in queue for each agent on phone).

Genesys performed sizing estimates with an assumed ratio of 10:1 for call center agents to supervisors.So, for example, if there are 1000 call center agents, the sizing is modeled based on an assumption of an additional 100 supervisor users as well.See Table 2 for additional details about how those Supervisor users were tested.

Note:

Queue monitoring is designed for IT, Management, and Team Lead personnel and not intended for all agents.If all agents must monitor ACD queues, scalability is reduced and Genesys recommends that you consult Sales Engineers for proper Session Manager sizing.

Table 2 Additional Details for Call Center Agents Scenario

Additional information

Scenario details

Supervisor clients configuration

  • 1 workgroup queue with default supervisor queue attributes. Change to a different queue every 5-10 minutes

  • 200 Agent statistics:

    • 20 Agents with 10 statistics each

    • The 10 statistics are the same for all agents however, each supervisor will monitor different stats randomly chosen from current period and current shift.

    • Watch all statistics on one workgroup. Change to different workgroup every 5-10 minutes.

  • One workgroup directory watch with user status. Change to different workgroup every 5-10 minutes.

  • 161 Alerts per supervisor

  • One Time In Status alert with On Call Status from 30-60 seconds on one workgroup

  • 100 Agent Statistic Alerts (20 agents X 5 stats X 1 range), non-time based

  • 60 Workgroup Statistic Alerts (3 workgroups X 10 stats X 2 ranges)

    • Approximately ten time-based and 50 non-time-based alerts

    • Send client memo to workgroup members on six of the statistics

Additional scenario details

  • 25 workgroups, 100 active members each, one workgroup per user

  • 20 CPS (All inbound ACD), recorded, with a max of 3800-4000 calls

  • On Call status and Follow Up enabled in the workgroups

  • Exchange 2010, Outlook Client 2010, Oracle 11gR2

Further, example scaling estimates were created based on tests performed with a particular set of hardware, as described in Table 3.Take into account individual variations from this example when determining an estimated sizing.

Table 3 Hardware Used for Sizing Estimates

Scalability Test Configuration

Function

Enterprise / Business Users

Call Center Agents

CIC server

CPU

Intel Xeon X5670

  • Dual 6-Core Hyperthreaded (24 Core)

  • 2.93GHz

CPU

Intel Xeon X5570

  • Dual Quad-Core Hyperthreaded (16 Core)

  • 2.93GHz

RAM

24GB

RAM

12GB

Off-Server Session Manager

CPU

Intel Xeon E5620

  • Quad-Core Hyperthreaded (8 Core)

  • 2.40GHz

CPU

Intel Xeon E5620

  • Quad-Core Hyperthreaded (8 Core)

  • 2.40GHz

RAM

6GB

RAM

6GB

The maximum load per Session Manager instance is shown in Table 4 in which the values indicate maximums where only one type of user exists at a site.However, actual installations have a mixture of business users and call center agents, resulting in a maximum load somewhere between the two values (see Table 5 for an explanation of how to calculate sizing with a mixture of user types). The maximum values refer to designing for the specified number of simultaneous users.If there are more users than the maximum for a Session Manager running on the CIC server, then the design must include sufficient off-server Session Manager machines to support all users.This guidance is because the on-server Session Manager instance does not accept connections if there are configured off-server Session Manager machines.For more information, see Configuring Session Manager Instances.

Table 4 Maximum Load per Session Manager

User Category

Session Manager (CIC Server)

Off-Server Session Manager

Total in Tested Scenario, Across All Off-Server Session Managers

Approximate maximum businessusers

5,000

5,000

10,000

Approximate maximum callcenteragents

2,500*

(+ 250 supervisors)

2,500*

(+ 250 supervisors)

4,000*

(+ 400 supervisors)

*Sizing estimates were performed with an assumed ratio of 10 to 1 for call center agents to supervisors.So, for example, if there are 2500 call center agents, the sizing is modeled based on an assumption of an additional 250 supervisor users as well.

When designing a large site for N+1 redundancy, the following equation can be used to determine the number of off-server Session Manager machines that are needed.The +1 adds an off-server Session Manager instance to provide resiliency for when another off-server Session Manager machine fails or is taken offline for maintenance. In addition, the +1 machine also aids in spreading the resource load while all the off-server Session Managers are functional.

Table 5 Sizing Calculation

Value

Definition

Number of off-server Session Manager Machines

Note:

Ceiling refers to rounding up to the next whole number.

Total number of simultaneously logged in business users.

Total number of simultaneously logged in call center agents.

Maximum number of simultaneous business users per off-server Session Manager machine.

This value is 5000, as per Table 4.

Maximum number of simultaneous call center agents per off-server Session Manager machine.

This value is 2500, as per Table 4.