The Anvil! is the hardware and redundancy architecture: the physical design that makes failure survivable. ScanCore is the software layer that decides what to do about it. As Alteeve’s platform evolves, ScanCore is increasingly the product itself, not a supporting feature of it. This article treats ScanCore as deserving its own explanation, separate from general Anvil! platform copy, for that reason.
ScanCore: The Decision Engine Behind The Anvil!
The Architecture Is the Anvil!. The Brains Is ScanCore.
What ScanCore Does
ScanCore is the rules-based decision engine at the core of the Anvil! platform. It continuously monitors hardware and software state across both nodes and runs weighted, additive scoring to compare simultaneous conditions, so it can judge which of several developing problems is the more dangerous one, not just detect that something is wrong.
That distinction matters because most availability platforms treat every fault the same way: a threshold trips, an alert fires. ScanCore doesn’t wait for a strict pass/fail line. A hard drive’s error counter incrementing from 1 to 4, for example, sets an alert on its own, well before that drive would cross a simple failure threshold. The system is reading a trend, not a switch.
This comparison isn’t a one-time judgment either. ScanCore continuously re-weighs conditions across both nodes, and its decisions can reverse as conditions change. If a workload is moved to Node A because Node B is degrading, and Node A subsequently starts failing worse than Node B did, ScanCore can move it back. The evaluation runs both ways, between both nodes, on an ongoing basis, not as a single one-way decision made and then forgotten.
What ScanCore Actually Controls
Where it counts, ScanCore doesn’t just observe. It orchestrates load shedding, shutdown, startup, and migration actions across the cluster, including shutting down an entire cluster if conditions call for it. The one decision it doesn’t carry out directly is fencing: ScanCore acts as a senior advisor to Pacemaker, but Pacemaker performs the actions itself. Across every one of these functions, ScanCore’s role is the same: it directs; the underlying components carry the decision out.
Deterministic, Not Probabilistic – and Why That’s a Strategic Choice
Because ScanCore’s logic is rules-based rather than probabilistic, its behaviour is fully deterministic: every state it can reach is predetermined by its scoring logic, built on real-world experience, not inferred.
This isn’t just a technical distinction from AI systems built on pattern matching, it’s a deliberate strategic choice. ScanCore often operates in conditions with no human present to catch a mistake in real time, particularly at the edge. A learning system that adjusts its own behaviour based on inferred patterns can arrive at a wrong conclusion with no one there to catch it before it acts. ScanCore’s rules mean there is no space between the rules and the outcome for something unpredictable to occur: the system cannot hallucinate, guess, or improvise, because every state it can reach was decided in advance by the people who built it, not inferred by the system itself in the moment. That is precisely the guarantee an unattended, unmonitored deployment requires.