Predictive maintenance scope
Applies to: Product direction; production coverage depends on a qualified release
Updated
Arcturus brings one reliability and management approach to individual machines, home clusters, fleets and data centers. Its predictive-maintenance scope starts with machine-level hardware conditions and extends to analytics across the data-center stack. A machine's condition can then be considered alongside the power, cooling, connectivity and facility conditions that affect its operation.
This page describes product direction. The production predictive capabilities described here remain in development. The public demonstrations currently use sample observations to show investigation, dependency review, incident records and maintenance planning.
Two levels of coverage#
| Arcturus scope | Intended environment | Predictive maintenance scope |
|---|---|---|
| Machine and home cluster | A personal PC or home cluster. | Machine-level hardware condition and maintenance needs, based on available observations from supported equipment. |
| Fleet and data center | Fleets, enterprise machines and data centers. | The same machine-level capability, extended to analytics across the surrounding data-center infrastructure and its dependencies. |
At the machine level, the work begins with the systems a person or team operates. Across a fleet or facility, the operating question extends to the connected services: which developing conditions could affect availability, what depends on the affected equipment, and where maintenance attention is needed.
The data-center stack#
| Domain | Intended coverage |
|---|---|
| Power | Utility supply, uninterruptible power supplies (UPS) and power distribution units (PDU). |
| Cooling | Chillers, computer room air handlers (CRAH) and coolant distribution units (CDU). |
| Networking | Switches, optical links and associated connectivity. |
| Compute and storage | Servers, storage systems and GPUs, including available machine-management observations. |
| Facility conditions | Fire, leaks and access-system observations relevant to the operating environment. |
Arcturus's intended value is the relationship between these domains. A server observation may need to be investigated alongside a cooling condition or a power event. Keeping the asset, observation time, dependency and affected service in context helps an operator determine what to investigate and plan the appropriate work.
The table is a product-scope map, not a compatibility list. Support for a particular controller, sensor, management system or vendor requires a documented integration and a qualified deployment. No vendor relationship is implied by an equipment category.
From an observation to maintenance work#
The intended workflow connects hardware and facility observations to an assessment of developing problems, their possible operational consequences and the maintenance work to consider. Operators need to retain the source and time of an observation, understand missing inputs, and distinguish an assessment from a confirmed diagnosis.
A planned maintenance action also needs an owner, an appropriate procedure and a record of what happened. Analytics, maintenance planning and work performed on equipment are separate capabilities. A predictive assessment alone does not establish that maintenance has been carried out or that a fault has been resolved.
What the public demonstrations show#
The data-center demonstration follows a facility condition through affected assets and a maintenance plan. The machine demonstration follows machine observations into incident work. They let a reader examine the operator workflow before evaluating a connected deployment.
These demonstrations do not connect to live equipment, issue a trained failure forecast or perform physical maintenance. For an evaluation, specify the equipment and observations available, the failure or maintenance question to investigate, and the outcome that would establish usefulness. The supplied release and evaluation scope should then identify which capability is actually being tested.