Cybee - Blog - News - The orchestrator, the keystone of the Cybee solution

The orchestrator, the keystone of the Cybee solution

Task orchestration is a key strategic component of modern cloud backup systems, where every resource (time, data, network) must be optimised.

Task orchestration: the key to intelligent, efficient Cloud backup

Task orchestration is a strategically essential component of modern Cloud backup systems, where every resource — time, data, network — must be optimised. This approach sits at the heart of Cybee’s backup orchestrator development.

In a modern Cloud backup system, every action matters. Every minute, every byte, every network resource is valuable. That is why task orchestration has become a strategic cornerstone of any high-performance backup solution. At Cybee, this topic is central to our thinking as we develop our backup orchestrator.

Why orchestrate tasks?

Backing up data is a start. Doing it at the right time, without conflicts, without overload, and with guaranteed restorability — that is orchestration.

The goal of orchestration is to intelligently schedule, prioritise, and execute tasks, taking into account all available resources and technical or business constraints.

In a distributed Cloud environment, orchestration becomes essential to:

  • Avoid conflicts between concurrent tasks
  • Prioritise critical backups
  • Respond to manual or automated events
  • Optimise resource utilisation (network, CPU, repository…)

Which tasks does orchestration cover?

The orchestrator is not limited to scheduled backups alone. It must also manage:

  • Manual backups triggered by a user
  • Immediate or scheduled restorations
  • Pruning (removal of obsolete data blocks)
  • Forget operations (deletion of snapshots)
  • Integrity verification
  • Agent updates
  • Repository maintenance

Each task type has its own constraints, priorities, and resource footprint.

Orchestration and resources: dynamic management

To operate effectively, the orchestrator must account for available resources:

  • LAN/WAN bandwidth
  • CPU & RAM (RAM considered secondary)
  • Available storage on repositories
  • Number of concurrent tasks per repository

These resources directly affect the ability to run multiple tasks in parallel. For instance, launching two large backups to the same repository can significantly degrade performance, or even cause failures.

Hence the need for intelligent scheduling based on:

  • Historical resource usage (throughput history)
  • The backup window
  • Estimated task duration
  • Repository locks

👉  One approach under consideration is the introduction of “constraint domains”, per client or per Cloud provider, to isolate prioritisation calculations.

Prioritisation: an adaptive scoring system

Not all tasks are equal. Some must be executed quickly; others can wait.

The envisaged system is based on dynamic scoring driven by:

  • Time elapsed since last execution
  • Remaining backup window
  • Estimated task duration
  • Resource cost
  • Resource locks
  • User priority (manual or automatic)

This logic allows the orchestrator to always select the best task to run right now, taking current conditions into account.

A concrete example: the new critical server

An administrator deploys a new critical server. They configure a backup plan and want to validate it immediately.

💡 In this scenario:

  • A manual task is launched with high priority
  • The orchestrator pauses or defers lower-priority tasks
  • It ensures the required resources are available
  • It temporarily locks the relevant repository

This scenario is typical of the priority user stories to be integrated from V1 of the system.

What’s next?

Cybee’s next steps:

  • Define resource metrics collection (throughput, duration, failure rate…)
  • Refine task scoring
  • Implement constraint domains
  • Manage coexistence of manual and scheduled tasks
  • Anticipate multi-tenant and multi-repository requirements

Task orchestration is not a luxury — it is an essential building block for ensuring the reliability, performance, and cyber-resilience of our Cloud backups.

Nos articles similaires

Tous nos articles