13.2 VMware vSphere, Extensions & Other Agentless Ingestion Paths

The official learning path's Ingestion lesson names three agentless routes explicitly: log ingestion, cloud integrations, and the VMware Monitor Hub. Section 13.1 covered the hyperscalers; this section covers the private-cloud and technology-specific paths that complete the Cloud Integrations and Ingestion portions of the blueprint.

---

VMware vSphere Monitoring

VMware Monitor Hub enables the ingestion of metrics and logs from VMware vSphere environments. It connects to vCenter to gather data on clusters, hosts, virtual machines, datastores, and disks, helping monitor health and performance.

The mechanics:

• An ActiveGate connects to the vCenter Server API over HTTPS using a read-only service account. No software is installed on ESXi hosts.
• Dynatrace discovers the vSphere inventory — datacenters, clusters, ESXi hosts, datastores, and virtual machines — and collects performance counters for each.
• Where a VM also runs OneAgent, Dynatrace correlates the guest to its hypervisor, so a monitored host entity knows which ESXi host it is running on.

That correlation is the exam-relevant payoff. Section 4.4 established that CPU steal time inside a guest proves the hypervisor is withholding cycles but cannot say why. With vSphere monitoring in place, the picture completes: you can see that the ESXi host is over-committed, identify the noisy neighbour VM sharing it, and act. Without it, the investigation stops at "something outside the VM."

Datastore metrics matter for the same reason at the storage layer. A datastore approaching capacity or exhibiting high latency degrades every VM placed on it, which is the classic explanation for a group of unrelated application hosts all slowing down simultaneously with no shared application dependency.

---

Extensions

Many technologies expose no place to install an agent — a network switch, a storage array, a mainframe subsystem, a third-party appliance. Extensions are the packaged collectors for these, and they are browsed, installed, and managed from the Dynatrace Hub (Section 1.1).

Common extension datasources:

Execution location is the detail exam questions target. Extensions run in one of two places:

• On the host, via OneAgent — for datasources local to a monitored machine, such as reading a local JMX endpoint or a local log-derived metric.
• Remotely, on an ActiveGate — for datasources polled over the network, such as SNMP devices or a remote SQL endpoint.

The consequence: an extension can only reach what its executing ActiveGate can reach. If a scenario describes SNMP devices in an isolated network segment, the answer is to place an ActiveGate inside that segment — the same network-zone reasoning as Section 2.3, applied to ingestion rather than to agent routing.

---

OpenTelemetry and Open Standards

Dynatrace ingests OpenTelemetry traces, metrics, and logs through the OTLP protocol, either directly to the platform endpoint or via an ActiveGate. This is the route for technologies OneAgent does not support, for teams with an existing OpenTelemetry investment, and for languages or runtimes outside the automatic-injection matrix. Chapter 12 covers OpenTelemetry and OpenPipeline in depth; the point here is that it belongs to the same agentless family — a way to get data in when automatic instrumentation is not an option.

---

The Generic Ingest APIs

When no extension, integration, or agent fits, Dynatrace exposes direct HTTP ingest endpoints:

• Metric ingest — line-protocol metric data points
• Log ingest — used in serverless environments or wherever OneAgent installation isn't possible
• Events ingest — deployment markers, configuration changes, custom annotations
• Business events ingest — JSON business records (Chapter 14)

These are the fallback of last resort in exam scenarios, and there is a cost dimension worth remembering: metric data points ingested this way are metered per data point (Section 1.4), so a poorly designed custom metric with high-cardinality dimensions can be dramatically more expensive than an equivalent extension.

---

Choosing the Right Ingestion Path

Work down this list and take the first match:

1. Can OneAgent be installed? Use it — it gives full-stack, code-level data automatically.
2. Is it a managed cloud service? Use the cloud integration for that provider (13.1).
3. Is it a hypervisor or virtual infrastructure? Use VMware vSphere monitoring.
4. Is there a Hub extension for the technology? Activate it, and place the executing ActiveGate where it can reach the target.
5. Does the workload already emit OpenTelemetry? Ingest OTLP.
6. None of the above? Use the generic ingest APIs, and design the data model carefully for cost.

Exam answers that skip straight to step 6 when step 1 or 4 would work are the most common trap in this topic.