Introduction

This Guide describes how the OnDemand Agents work for Apica LoadTest (ALT) and tips not to overload them.

The typical subscription for an ALT customer is “OnDemand.”

Locations

Location Map	OnDemand Agent Locations

Location Map

OnDemand Agent Locations

When you set up a load test, you choose from a set of locations from where you want to run your load test:

Each location has several available OnDemand Agents.

OnDemand Agents

On the following link, you can find up-to-date information about locations and agents:

ALT Agent Global Locations with IPs

The current number of available agents per location:

Location	Available agents
Australia - Sydney	10
Brazil - Sao Paolo	10
Canada - Montreal	10
Germany - Frankfurt	10
France - Paris	50
Ireland - Dublin	10
India - Mumbai	10
Japan - Tokyo	40
South Korea - Seoul	10
Sweden - Stockholm	50
Singapore	10
UK - London	10
USA - Ohio	250
USA - Oregon	250
USA - San Francisco	250
USA - Washington DC	250

Total Agents	1220

How Customer Jobs are Allocated

Each OnDemand Agent has a limit of 500-1,000 Virtual Users, depending on the location and contract.

When a customer runs a load test job, one or more agents are:

Reserved for that customer only
- For each location
- For the duration of the job
- For a maximum of 500 Virtual Users (VU) per Agent
The allocation takes place before the jobs start.
These agents are not available for any other customers or jobs for the duration of the test.

Example:

A customer wants to run a load test of 1,800 VU’s from 2 locations. The system will then allocate 2 agents in each location, and each agent will run 450 VU’s.

When using Performance Test Scenarios and several tests, each test/Job is handled separately.

Performance of a Load test Agent

As before, each Agent is allocated up to 500 VU’s, but the CPU load on the Agents can vary a lot depending on the script:

Each of these conditions calls for More/Higher CPUs:

The number of calls per “page.”
- Higher calls
If calls are made in “parallel” or “serial.”
- Parallel Calls
The think time (pause) between each “page.”
- Shorter Think Times
The response times of the called resources
- Shorter Response Time

When an Agent is overloaded (e.g., CPU > 90%), it will limit the throughput, and the returned performance data will be wrong (too low).

In general, an Agent can handle up to 1,000 calls per second.

Can you see if an Agent is overloaded?

1. Compare Response Times

Loop Time > Page Times = Agent Overload

The typical symptom of an overloaded Agent is that the total duration of a test iteration (“loop time”) is clearly longer than the sum of the “page times.”

This is because not all execution time is included in the page times. For instance, SSL handshakes, HTTP handshakes, and garbage collection.

When the Agent's CPU is overused, these additional processes may take significant time for the agent to process so that they will be added to the Loop time but NOT the Page Time.

Example:

On the test “Overview” tab, check the Average Response time Per Loop	Compare the Loop Response time to the Page Summary Total
	There was no difference; the Loop included each of the 5-page response times, so the Agent worked fine!

Looking at the detailed result file

Looking at the detailed result file
Once a job is finished, you can download the detailed result file and analyze it. In the “Jobs” view, click the download icon for the job you’re interested in.
This will get you an xxx.prxres file. Put that file in the ZebraTester/MyTests folder. Open the file in the ZebraTester UI/Project Navigator:
Then click the magnifying glass next to the Load Test name:
Then select the Diagram: Load Generator Performance:
Finally, look at the CPU Usage graph:	In this example, the usage is very low, so we’re fine!

2. Leverage SSL Cache Efficiencies

More SSL Cache Hits Help Agent CPU

Check SSL Cache Efficiency and Percent time Spent on SSL.

Check SSL Cache Efficiency and Percent time Spent on SSL.
You can check for this in the result file (see the previous section) by selecting Diagram: SSL Cache Efficiency. The example below shows that a new SSL handshake was performed for every request. This can be CPU intensive, so keep the SSL Cache-hits as high as possible; Offloading this to the cache keeps the Agent CPU lower.
Also, look at the Results per Page \| URL (Details) view. Compare the numbers in the Av SSL/TLS column with the Av Time column. In the example below, about 10% of the time is spent on SSL.

Apica LoadTest Documentation

ALT OnDemand Agents Guide

Introduction

Locations

OnDemand Agents

How Customer Jobs are Allocated

Example:

Performance of a Load test Agent

Can you see if an Agent is overloaded?

1. Compare Response Times

Loop Time > Page Times = Agent Overload

Example:

Looking at the detailed result file

Looking at the detailed result file

2. Leverage SSL Cache Efficiencies

More SSL Cache Hits Help Agent CPU

Check SSL Cache Efficiency and Percent time Spent on SSL.

Check SSL Cache Efficiency and Percent time Spent on SSL.