Resource allocation considerations

ArcGIS Velocity can be resource-intensive when ingesting large volumes of events per second, performing complex real-time analyses, or both. Because Velocity offers broad capabilities and solves real-time use cases across nearly all industries, it is recommended to deploy Velocity on a machine that aligns with your specific use cases. Different use cases require different machine resources.

Below are general guidelines for selecting machine hardware that is appropriate for a Velocity deployment. It is recommended that you meet at least the minimum system requirements when getting started with Velocity.

CPU

In Velocity, the machine CPU is used for real-time feature processing, analysis, and dissemination. The speed of the machine’s CPU directly correlates to the number of features per second that Velocity can process. The greater the number of bytes per second you are receiving, the more CPU capacity is required.

RAM

Velocity uses machine memory (RAM) to manage feature caches during real-time analyses, store track ID-based features, or cache features as geofences. The capacity for these tasks in Velocity is directly related to the amount of RAM dedicated on the machine. The available RAM on a Velocity machine ultimately determines how many track-based features can be stored, or how many geometry vertices can be retained for analysis. The speed at which geofences can be accessed is directly correlated to the type and speed of the RAM. If Velocity is used to perform complex geofence or track-based operations with its real-time analytic tools, consider choosing increased or performance-optimized RAM.

Network

In a production environment, Velocity relies on external connections to receive data. Consider increasing network bandwidth and reducing latency when the ingestion of a large volume of data is expected. Large in this case means data with many attributes, a high number of features in each payload, or high frequency updates.

Local high-performance disk

When deploying Velocity, select an optimal location for storing the configuration and message queues. It is recommended that you use a local storage device optimized for handling small, random input/output (I/O) operations.

Storage optimized for large, sequential reads and writes, such as those commonly used for imagery and video, is not suitable for Velocity. If your file storage mechanism does not handle small, random input/output (I/O) operations efficiently, you may experience increased data processing times or data queue overload.