Bot deployment and concurrent operations

List of maximum concurrent operations and estimated deployment times.

Automation Anywhere supports up-to 1000 simultaneous bot deployments and executions.

Concurrent deployment operations

Note: The entity count in the following list provides information about the maximum number of entities that the Automation Anywhere Enterprise can support. Deploying additional Control Rooms or machine resources does not affect the entity count. The time taken for deployment depends on the network quality. An increase in the number of Control Rooms and the network connectivity speed improves the throughput and enables faster deployment.
Entity Types and Counts
Entity Type Count
Users 5000
Roles 2000
Schedules 1500
Audit entries 50000
Lockers 100
Credentials 5000
Repository files 2500
Repository folders 1250
Queues 10
Device pools 10
Work items 1,000,000
Concurrent deployment operations for unattended bot
Action Maximum with one Control Room Maximum with two Control Rooms with High Availability (HA) Configuration Maximum with three Control Rooms with High Availability (HA) Configuration Maximum with four Control Rooms with High Availability (HA) Configuration
Deployment and execution 1000 1000 1500 2000
Repository upload from Bot Creator 100 100 150 200
Repository download from Bot Creator 100 100 150 200
Note: For more information on High Availability (HA) deployment model, see Enterprise 11: High Availability deployment model.
Concurrent deployment operations for attended bot
Action Maximum with one Control Room Maximum with two Control Rooms with High Availability (HA) Configuration Maximum with three Control Rooms with High Availability (HA) Configuration Maximum with four Control Rooms with High Availability (HA) Configuration
Connected attended device 4500 4500 8000 12000
Deployment time estimates
Number of devices bot and dependencies total size Approximate deployment time
1000 10 MB 1 minute
1000 50 MB 1 minute
1000 100 MB 5 minutes

Devices and the Control Room coordinate to implement a fair queuing strategy for download and uploads of chunked data to the repository.

As the number of devices simultaneously downloading and uploading increase, the time taken for the Control Room to start processing a chunk request increases. A device waits for up to two minutes for a response before timing out a request to upload or download. With the default limit of 10 parallel processed repository requests and simultaneous deployment and execution to 1000 devices, the average time to queue and process a chunk is approximately 10 seconds.

Note: Downloads and uploads from bot creators are subject to the same QoS policies as deployments.

Tuning parallel repository operations

If the network connectivity between devices and the Control Room is slower than the reference configuration, or the CPU is under-utilized during deployment, the number of parallel repository requests can be increased for better throughput.

If the CPU speed of the Control Room is lower than the reference configuration or network speed is higher it might be necessary to lower the number of parallel repository requests allowed to avoid loss of availability due to excessive CPU utilization throughout deployment.

If during execution the Control Room or database server experience a loss of availability due to excess CPU usage, the default of 10 parallel processed requests can be adjusted down.

To adjust the number of parallel repository requests:

  1. Create a boot.server.properties file in the Program Files\Automation Anywhere\Enterprise\config directory if not already present.
  2. Add the line:

    requests.max.client.concurrent.repo=<number>