The Elastic Collector Profile is a dynamic resource management system designed to optimize the registration and operation of gateways with the OpsRamp cloud. Unlike default management profiles, elastic collector profiles enable systems to handle applications and integrations with a high number of resources by:

  • Scaling Up and Scaling Down: Dynamically adjusting the number of profiles to accommodate varying resource demands.
  • Rebalancing Resources: Distributing resources across dynamically created profiles (replica profiles) to ensure efficient load distribution and system stability.

This document outlines the design, functionality, and benefits of elastic management profiles, highlighting their role in enabling scalable and efficient resource management.

Permissions

To enable this feature, please contact our OpsRamp support team to activate the feature flag.

Current Design Limitations

In traditional setups, default collector profiles (Profile type=standard) are responsible for managing resources tied to applications and integrations. However, when an application has a significantly high number of resources, the following limitations arise:

  • Overburdened Profiles: A single management profile cannot efficiently handle all resources, leading to performance bottlenecks.
  • Lack of Scalability: Default profiles cannot scale dynamically to accommodate fluctuating resource requirements.

How Elastic Collector Profiles Address these Challenges?

Elastic collector profiles overcome these limitations by introducing scaling and load balancing using a new profile type such as Elastic:

  1. Scale-Up and Scale-Down Operations:
  • When the number of resources exceeds the capacity of a single collector profile:
    • Scale-Up: A new replica profile is dynamically created to handle the additional resources, with resources evenly distributed across the original profile and the newly created replica profile.
  • During periods of low demand:
    • Scale-Down: Unused replica profiles are deleted, and their resources are redistributed.