Edge and distributed deployments under audit.

Distribution multiplies the measurement problem

A centralized data center is straightforward to instrument. A distributed estate, with software running at retail sites, factory floors, remote offices or edge appliances, is not. Each location is a place where an ILMT agent has to be installed, kept running, and reported on, and each is a place where that can quietly fail. The sub-capacity rules do not soften for distance. The same conditions hold everywhere: ILMT or an approved tool deployed within 90 days of first eligible deployment, running continuously, with quarterly reports retained for two years. Fail any of them at any location and IBM defaults to full-capacity charging for that node.

The risk is structural. Central IT often has clean visibility into the core estate and patchy visibility into the edge, where servers are commissioned locally, networks are intermittent, and nobody owns the inventory tooling. That gap is invisible day to day and very visible the moment an audit asks for continuous measurement of every install.

The patterns that surface under audit

• Edge nodes never enrolled. Software deployed to a remote site that was never added to ILMT inventory has no sub-capacity evidence at all, so the count defaults to the full capacity of that hardware.
• Intermittent connectivity breaks reporting. Edge appliances that connect only occasionally can miss the continuous measurement the rules require, leaving gaps in the quarterly record.
• Local provisioning outside the standard build. A site that stands up its own server bypasses the central image that would have carried the agent, and the install goes uncounted.
• Counting the wrong capacity. On small edge hardware the difference between virtual cores used and full physical capacity may be modest per node, but multiplied across many sites the full-capacity default becomes a large aggregate exposure.

Bringing a distributed estate back into a defensible position

The work starts with discovery: an honest map of where IBM software actually runs, including the locations central tooling never reached. From there it is the familiar sub-capacity discipline applied at scale, confirming every node runs on eligible technology, is enrolled in ILMT, and is measured continuously. Where history has gaps, the position is reconstructed from whatever evidence exists and the full-capacity assumption is challenged node by node. A distributed estate is not inherently non-compliant; it is just harder to prove, and the proof is what an audit demands.