IBM Instana Observability Administrator - Professional key terms

A unique identifier issued by the Instana backend that authorizes an agent to register with a specific tenant. It is supplied during installation and can be rotated, so a compromised key can be replaced without reinstalling every host.

An Instana agent installed as a system service directly on a host, managed by the operating system's service manager. It monitors the host and its local processes continuously and survives reboots, unlike a containerized agent tied to a container lifecycle.

An agent deployed as a container or short-lived workload, typically through a Kubernetes DaemonSet or operator, rather than as a host service. It is created and destroyed with the cluster and is upgraded by changing the deployment specification.

A Kubernetes controller that installs, configures, and upgrades Instana agents across a cluster from a single custom resource. It reconciles agent DaemonSets and configuration automatically, removing the need to patch each node individually.

The central YAML file on an agent host that declares the backend endpoint, agent key, enabled sensors, and per-sensor settings. Changes are picked up on restart or reload, and the file is the primary artifact for troubleshooting agent behavior.

The hostname and port an agent uses to reach the Instana backend, differing by region and by SaaS versus self-hosted deployment. An incorrect or unreachable endpoint prevents registration and is a common cause of agents showing as offline.

A plugin inside the Instana agent that detects a specific technology, such as a JVM, database, or middleware server, and collects its metrics and configuration. Sensors are enabled or disabled per technology and can be tuned with technology-specific settings.

The agent's process of scanning running processes and open ports to decide which sensors should activate. Detection results explain why a sensor is or is not collecting data for a given host and are visible in the agent's diagnostic output.

A backend-side rule that controls which agent and sensor versions a group of agents should run, and when they may update. It lets administrators stage rollouts by host group instead of upgrading every agent at once.

A periodic verification that an agent is running, connected to the backend, and reporting recent data. Failed checks surface as offline or stale agents and usually point to network, credential, or resource problems on the host.

The collected logs, configuration dump, and connectivity test output an agent can produce on demand to support troubleshooting. Support engineers use this bundle to determine whether failures are local, network-related, or backend-related.

A saved or ad hoc query that filters infrastructure and application data by tag, such as host zone, service name, or Kubernetes namespace. It scopes dashboards, alerts, and analyses to a precise subset of the monitored estate.

A key-value attribute attached to monitored entities, either discovered automatically or defined by an administrator. Tags are the vocabulary of Dynamic Focus Queries and drive grouping, filtering, and alert scoping across the platform.

A logical grouping of services and endpoints that represents one application for monitoring purposes. It defines the boundary within which Instana builds the service map, calculates health, and correlates traces and calls.

A visual graph of the services in an application perspective and the calls between them, annotated with latency and error rates. It helps administrators spot unhealthy dependencies and understand traffic flow during incidents.

The complete record of a single request as it travels through multiple services, composed of timed spans with parent-child relationships. Traces reveal where latency accumulates and which call in a chain returned an error.

One timed operation within a trace, such as a database query or an outbound HTTP call, with a start time, duration, and status. Spans nest to form the call tree that shows exactly where time was spent.

The rule set that decides which requests are recorded as full traces rather than only aggregated metrics. Administrators tune sampling to keep visibility of errors and slow calls without overwhelming backend storage.

The code-level or agent-level mechanism that captures spans for an application, either automatically for supported runtimes or manually through an SDK. Correct instrumentation is required before traces and service maps appear.

An alert definition that watches a metric for a defined entity scope and fires when the value crosses a threshold or behaves abnormally. It combines a Dynamic Focus Query, a metric, a condition, and one or more notification channels.

A configured destination such as email, Slack, PagerDuty, or a generic webhook that receives alert notifications. Channels are defined once and reused across many alerts, and each can carry its own routing and payload settings.

A recorded occurrence in the platform, such as an alert firing, a deployment, or an infrastructure change. Events provide the timeline used to correlate incidents with changes and to drive issue escalation.

A grouped set of related events and alerts that together describe a service problem requiring attention. Incidents carry state, ownership, and timeline so responders can coordinate and record resolution.

A scheduled period during which alerts are suppressed or treated differently for a chosen scope, such as hosts being patched. It prevents planned work from generating noise while still recording events for later review.

A rotation that defines which responders receive notifications for an alert at a given time. It routes escalations to the right person or team and is referenced by alert channels when paging is required.

The permission model in which users receive roles, and roles grant specific capabilities over specific scopes such as applications or infrastructure. It lets administrators grant least-privilege access without managing per-user permissions.

A grouping of users that owns resources and receives access assignments as a unit. Teams simplify administration because permissions and alert routing can be granted once to the group rather than to each member.

The set of entities, such as applications, websites, or infrastructure groups, to which a role assignment applies. A user may hold one role globally and a different role within a narrower scope.

An authentication setup in which users log in through an external identity provider using a standard protocol instead of a local password. It centralizes account lifecycle and enforces the organization's authentication policy.

The signed statement from an identity provider that tells the platform who the user is and which groups they belong to. Group claims are commonly mapped to teams and roles during single sign-on configuration.

A credential used by scripts and integrations to call the platform's APIs without an interactive login. Tokens carry the permissions of their owner or assigned role and should be scoped narrowly and rotated regularly.

The policy that determines how long different classes of monitoring data, such as metrics, traces, and events, remain queryable. Retention settings balance troubleshooting needs against storage cost and regulatory requirements.

Configuration that removes or masks sensitive values, such as user identifiers or query parameters, before data is stored. It supports privacy compliance by preventing personal information from persisting in traces and logs.

The agent component responsible for host-level metrics such as CPU, memory, disk, and network, plus process discovery. It is the foundation for infrastructure monitoring and for activating technology sensors on that machine.

Collection of cluster, node, namespace, workload, and pod metrics and events through an agent deployed into the cluster. It links container performance to the underlying hosts and to the applications running inside them.

A saved grouping of hosts or containers based on tags, used to organize the estate and scope access or alerts. Groups make large environments navigable and support consistent filtering across views.

A local URL an agent or backend exposes to report its own readiness and dependency status. Operators and load balancers poll it to detect degraded components before users notice missing data.