Glossary

ACM is the abbreviation for the AWS Certificate Manager, which creates, stores, and renews public and private SSL/TLS X.509 certificates and keys that protect AWS websites and applications.

An entity that collects logs or other metrics and sends them to the Kloudfuse Platform.

Aggregation simplifies a large dataset into key values such as average, sum, or count. It summarizes data to reduce processing and storage costs.

An alert signals a change in system health that may indicate a problem. It generates a notification so operators can investigate and take corrective action.

Metadata attached to Kubernetes objects. Use annotations to store non-identifying data that tools and systems can read.

Anomaly detection finds unexpected patterns in telemetry data. It highlights issues that need investigation.

The Application Performance Index, a single metric that represents the service quality on the scale of 0 to 1. Higher values indicate better performance, while lower values report low quality. APDEX combines the metrics of RED into a single measurement.

An acronym for Application Performance Monitoring, APM tracks and monitors application performance metrics like response times, error rates, and transaction traces. APM identifies and addresses issues that impact the application’s performance and user experience.

A set of public methods and properties that enables users to retrieve and modify objects through REST endpoints.

APM metric alerts are similar to metric alerts; with controls tailored specifically to APM. Use alerts at the service level on hits, errors, and a variety of latency measures.

The Application Performance Index, a single metric that represents the service quality on the scale of 0 to 1. Higher values indicate better performance, while lower values report low quality. APDEX combines the metrics of RED into a single measurement.

An acronym for Advanced Service Monitoring, ASM consists of technologies that help manage and monitor distributed applications within a cloud environment, and provide centralized visibility into their health and performance. ASM enhances observability by providing a unified view of complex microservice interactions through centralized management, traffic management, and observability data gathering.

An audit trail records all actions and changes in a system. It provides traceability for accountability and debugging.

Autoscaling automatically increases or decreases application resources in response to demand. It ensures performance and cost-efficiency.

Availability is the system’s ability to consistently respond to user requests, even if some individual nodes or components are experiencing failures.

A distinct and isolated data center in a cloud provider’s infrastructure, designed to provide fault tolerance and high availability for cloud-based services.

Capacity planning forecasts and manages the resources an application needs to handle current and future loads.

Cardinality of a data attribute is the number of possible distinct values, or uniqueness, it can have. See also High cardinality and Low cardinality

A chart represents data or information in a graphical format. Most charts in Kloudfuse are line graphs, stacked bar graphs, stacked area graphs, or point graphs. Terms “chart” and “graph” are interchangeable.

In Kloudfuse, a group of filters that identify the attributes of cloud-based services.

Unique identifier for a user or organization’s account in a public cloud provider’s system.

The process of restoring archived logs into queryable storage. Use it for investigating older issues.

Consistency is the implementation of consistent user interfaces and interaction patterns across different parts of the distributed system, regardless of which component is currently handling the request.

In Kloudfuse, a group of filters that identify the attributes of cloud-based container-based virtualizations. Includes container id.

Unique identifier for a container.

A cumulative metric that represents a single monotonically increasing counter, where the value can only increase, or be reset to zero on restart.

An extension of the Kubernetes API. Used to define and manage custom configurations like ClusterLogging and ClusterLogForwarder.

A data lake stores logs, metrics, traces, and events in one place so you can easily correlate and analyze them.

Dimensionality refers to the number of attributes in a dataset. Higher dimensionality increases data complexity.

A documented strategy, abbreviated as DRP, that outlines how an organization can quickly resume its critical IT operations using backup systems and procedures to minimize downtime and data loss after a disruptive event — a natural disaster, power outage, cyber attack, or hardware failure.

The name of a container within a system or application; used for display purposes or identification.

Distributed tracing follows a request across multiple services to pinpoint where issues occur.

Duration of a span is the difference between the span end time and the span start time. Contrast with execution time.

Extended Berkeley Packet Filter (eBPF) is a Linux kernel technology that enables users to run programs in a protected environment within the kernel. eBPF programs loaded into the kernel using the bpf(2) syscall. The user provides them as a binary blobs of eBPF machine instructions. eBPF allows you to:

Represents the environment or deployment stage ( development, staging, production) of a system or application.

Similar to env, represents the environment or deployment context.

An event is something that has happened in a system at a point in time. Events are discrete occurrences with precise temporal and numerical values. Through events, we can track crucial indidences, and detect potential problems associated with user requests. Because events are very time-sensitive, they include timestamps.

Total time that a span is active , not waiting for a child span to complete, scaled according to the number of concurrent active spans.

An exporter collects metrics from a system and converts them into a format that observability platforms like Kloudfuse can process.

Attributes that the ingester extracts from a log line.

An automatically-detercted structure of a log line.

A flame graph is a visual representation of profiling data that analyzes call stacks and identifies performance hotspots.

A log collector that runs on every node. Collects logs from containers and system services and forwards them to a defined output like Elasticsearch or S3.

Kloudfuse developed FuseQL as a query language for searching across logs data for a range of applications. It has flexible parameters for answering highly complex questions. See FuseQL.

A metric that represents a single numerical value that can increase or decrease.

This query language fetches data from multiple data sources with a single API call. See GraphQL.

gRPC is a high-performance, open-source framework for remote procedure calls (RPCs). gRPC enables client applications to call methods on a server application on a different machine as if it were local, making it easier to create distributed applications and services.

High cardinality is a condition when individual attributes within a dataset have a large number of unique values, compared to the total number of entries. We also refer to high cardinality when combinations of dimensions lead to a large number of unique permutations.

HIPAA is the accepted abbreviation for the Health Insurance Portability and Accountability Act of 1996. It is the United States federal law that aims to protect patient health information, and also ensure that the insurance coverage is portable. It sets national standards for the protection of Protected Health Information (PHI).

A metric that tracks the distribution of observed values. A histogram samples observations (request durations, response sizes, and so on) and counts them in configurable buckets. It also provides a sum of all observed values.

An accronym for Hypertext Transfer Protocol, http is a set of rules that govern how to exchange information between a client and a web server.

Organizes log data into searchable structures. Helps speed up queries by mapping data to indexed fields.

Infrastructure is the foundational layer, or backbone, of hardware, software, and resources that support the operation, deployment, and scalability of software applications and services. It includes a range of components and technologies that deliver computing, networking, storage, and other services for running and maintaining software systems.

A predefined configuration of virtual hardware resources in a cloud provider’s infrastructure.

Instrumentation adds code that captures performance and telemetry data from your app.

A feature in Cloud Observability that reports how complete and consistent telemetry data is for a given service.

Automatically creates a user account in Cloud Observability the first time someone logs in with Identity Provider (IdP) credentials. Assigns a default role during provisioning.

A key-value store is a NoSQL database that stores data as pairs for quick access and updates.

A Kubernetes resource that creates jobs that run at specified intervals, defined by a cronological schedule. Performs regular scheduled actions such as backups, report generation, and so on.

The name of a container that runs within a Kubernetes pod, a basic building block of Kubernetes applications.

The name of a Kubernetes cluster, a group of nodes that run containerized applications managed by Kubernetes.

A Kubernetes resource that ensures a specific pod runs on all or selected nodes within a cluster.

A Kubernetes resource that represents a set of identical pods, ensuring application scalability and fault tolerance.

A Kubernetes resource that manages batch tasks or processes that run to completion.

A logical partition or virtual cluster that isolates and organizes resources within a Kubernetes cluster.

A worker machine within a Kubernetes cluster, responsible for running containers and managing their lifecycle.

A Kubernetes resource that ensures a specified number of replicas (identical pods) are running at all times.

A method for exposing a network application that runs as one or more pods in your cluster.

Runs a group of pods and maintains a "sticky" identity for each; useful for managing applications that need persistent storage or a stable, unique network identity.

In Kloudfuse, a group of filters that identify the attributes of services running in Kubernetes.

The core service that validates and manages Kubernetes API requests. Acts as the gateway for controlling cluster state.

In Kloudfuse, a group of filters that identify the programming language of the integrated telemetry software development kit (SDK), including go, nodejs, cpp, java, dotnet, rust, python, php, and ruby.

Latency is the time taken for a single operation to complete, often measured as response time.

Lead time for changes tracks how long it takes code to go from commit to production.

Load balancing spreads traffic across servers to improve reliability and performance.

Log aggregation collects logs from many sources and stores them in one place for analysis.

A discrete log entry, with associated information: log line, log labels, facets, and fingerprint.

The key/value pairs associated with the log line.

The severity of a log event - info, error, warning, trace, debug, or notice.

This is the emitted log message. In practical terms, the maximum size of a log line is 1 MB.
See log event, log labels, facets, and fingerprint.

Log management organizes, indexes, and analyzes logs to help you troubleshoot and monitor systems.

Moves logs from cold storage to hot storage for querying and analysis. Useful for investigating historical issues without always paying for high-cost storage.

LogQL is Grafana Loki’s PromQL-inspired query language for searching across logs. See LogQL.

Logs record all activities that occur within your system; they are a history of the system’s behavior during a specific time interval. Logs are essential for successful debugging. When we parse log data, we can develop an insight into application performance that we cannot determine through APIs or application databases.

This conditions describes situations of low counts of unique values in the dataset, compared to the total number of entries.

An acronym for Metrics, Events, Logs, and Traces, MELT is a framework that provides insights into system performance, health, and behavior. MELT can help teams quickly identify, diagnose, and resolve issues, while optimizing system performance.

The number of unique metric time series from the combination of metric name and dimensions.

Metrics are numerical measurements that provide an insight into a system’s performance. Use metrics, such as error rate and CPU% use, in mathematical modeling and forecasting to represent specific data structures.

A node is a single machine or virtual instance in a network or cluster that performs work and runs services or containers.

Noise is the set of irrelevant or low-priority alerts that distract from real issues. Reducing noise helps teams focus on what actually needs attention.

O11y is shorthand for “observability.” It helps you understand your system using external signals like logs, metrics, and traces.

An observability platform aggregates and visualizes telemetric data collected from application and infrastructure components in a distributed environment. It monitors and analyzes application behavior, transactional data, and the various types of infrastructure that support application delivery, making it possible to proactively address issues before they become serious concerns.

OTLP(OpenTelemetry Protocol) defines how telemetry data moves between systems, collectors, and observability platforms.

Performance is the efficiency and effectiveness of the system in handling workloads by distributing tasks across multiple nodes. It is one of the key aspects of usability.

The PostgreSQL network protocol that governs how clients and servers interact in a PostgreSQL environment.

The smallest deployable unit of computing that you can create and manage in Kubernetes.

The name of a specific pod, representing a single instance of a running process in a Kubernetes cluster.

A named, isolated environment within a cloud account for organizing and managing resources.

Prometheus Query Language enables you to select and aggregate metric time series data in real time. See PromQL.

Protocol refers to the methodology for network communication, such as http or pg.

Shortened form of Protocol Buffers, protobuf is an efficient, language-agnostic data serialization mechanism for defining structured data in a . proto file. You can use it to generate source code that writes and reads data from different data streams.

A query is a request or command that retrieves, manipulates, or manages data stored in a database or information system.

Represents the type or category of a query performed on a database or data store, such as SELECT, INSERT, UPDATE, or DELETE.

Raw data is source data that has not been transformed, organized, or analysed; it is in its original form.

Refers to RED Metrics: and acronym for Requests, Errors, and Duration (Latency).

Redundancy means having backup components in place to ensure service continues even if a part fails. This improves reliability and fault tolerance.

A geographical location that hosts a cloud provider’s data centers.

Reliability is the ability of a system, application, or service to consistently perform its intended functions under specific conditions for a specified period. Along with availability and usability, reliability is a key attribute of quality in software and services.

Resilience is the ability of a system or application to gracefully and effectively handle unexpected failures, disruptions, errors, or adverse conditions while maintaining core functionality and minimizing the impact on users and operations. It is one of the key aspects of usability.

Resources represent a particular domain of a customer application - they are typically an instrumented web endpoint, database query, or background job.

An aggregation of data points over time, such as a weekly average or percentile.

The maximum amount of data the organization can tolerate losing; the goal an organization sets for the maximum length of time it can take to restore normal operations following an outage or data loss.

The maximum amount of time that it is acceptable to wait to restore a system or service after a disruption.

RUM is an acronym for Real User Monitoring, a technology that measures and tracks the end-user experience of an application or website. It is a core feature of APM, and a key component for gaining real-world observability into web performance and service availability.

In sampling, we programmatically select a subset or representative group of data points or items from a larger dataset for analysis, testing, or inspection.

Scalability is the ability to handle increasing workloads by adding more nodes to the system without significant performance degradation.

Pre-aggregated datasets that Kloudfuse generates at scheduled intervals to improve query performance and efficiency. Instead of expensive real-time queries on raw data, scheduled views store advance results, enabling faster access to summarized information.

Services are the building blocks of modern microservice architectures - broadly, a service groups together endpoints, queries, or jobs to build your application.

The Service Map is a graphical workflow representation of all services in the deployment.

A service entry span records the entry point method for a request to a service. Kloudfuse APM interface shows service entry spans when the immediate parent on a flame graph is a different color than the span.

Subdivisions of an Elasticsearch index. Improve performance by parallelizing data storage and query execution.

SLAs define the level of service and performance standards that the service provider is expected to meet. SLAs serve as a means of establishing clear expectations, responsibilities, and consequences for service quality and delivery.

Acronym for Service Level Indicator, this metric measures how well a service meets customer expectations. SLIs gauge the reliability of a service, and report the percentage of the service effectiveness. To calculate the SLI, divide the number of good events by the total number of valid events, and then multiply by 100.

SLOs are critical components of SLAs. They establish clear, measurable expectations for service providers and stakeholders, help to ensure that the service meets the needs of its users, and align with business or operational requirements. The best SLOs set a minimum standard for performance, are chosen thoughtfully with business objectives in mind, and focus only on measurable metrics.

The Service Map is a graphical workflow representation of all services in the deployment.

Spans are integral parts of a distributed system, and the basic element in distributed tracing. A span represents a logical unit of work in a distributed system for a specified period. It is a single operation within a trace. Multiple spans construct a trace.

You can tag spans in APM, as key-value pairs, to correlate requests and filters.

In Kloudfuse, a group of filters that identify the environment where the span runs. Includes http, grpc, db, queue, and custom.

The source application that emits logs or other metrics.

Synthetic monitoring uses scripted tests that simulate user interactions with an app to measure performance and availability before users encounter issues.

Refers to the name of a specific table within a database or data store where data is organized and stored.

A label on a node that restricts which pods can run on it. Prevents unwanted scheduling.

Telemetry data is automatically collected and transmitted from remote or inaccessible sources to a centralized location for monitoring and analysis. Metrics, events, logs, and traces each provide crucial insights into the application’s performance, latency, throughput, and resource utilization.

The telemetry type is the specific category or classification of data collected through telemetry, such as logs, events, distributed tracing, and so on.

A threshold is a defined limit (such as CPU usage > 80%) that, when exceeded, triggers an alert or action.

Throughput is the number of operations or requests a system can process per unit of time.

Time series is a set of data points indexed in chronological time order. Commonly, a time series is a sequence of successive equally-spaced points in time, or a sequence of discrete-time data.

In DBSCAN, the tolerance level, or eps, determines the clustering radius of the neighborhood around each point. The eps controls the sensitivity of outlier detection. A lower tolerance detects more subtle outliers, while a higher tolerance detects only the most significant deviations.

A trace is a collection of operations that represents a unique transaction handled by an application and its constituent services. It is the path of a request or workflow as it progresses from one component of the system to another, capturing the entire request flow through a distributed system.

Trace metrics identify and alert on hits, errors, or latency. Kloudfuse automatically collects trace metrics and retains them for 15 months, similar to other Kloudfuse metrics.

A root span tracks the entry point method for the trace. Its start marks the beginning of the trace.

TraceQL is Grafana’s query language that selects traces. See TraceQL.

A trace is a collection of operations that represents a unique transaction handled by an application and its constituent services. It is the path of a request or workflow as it progresses from one component of the system to another, capturing the entire request flow through a distributed system.

The user should not have to be aware of the system’s distributed nature, and should be able to interact with it as if it were a single, centralized system.

Usability is a measure of how easy and intuitive it is for users to interact with a distributed system, without needing to understand the complex underlying architecture or the individual components.

The USE method for monitoring system performance focuses on three key metrics: Utilization, Saturation, and Errors.

Vertical scaling means increasing a system’s capacity by adding more power (CPU, RAM) to a single server or node.

Visualization is the graphical representation of data or information. It means that we creatie visual representations (charts, graphs, diagrams, maps, and dashboards) to make complex data more understandable, accessible, and interpretable.

A workload is the amount and type of processing done by an application or service. It includes user requests, background jobs, and system tasks.

XML logs are structured log files formatted using XML, often used in legacy systems. They’re machine-readable, but harder to query than JSON.

Zone awareness is a deployment strategy that distributes workloads across multiple availability zones to improve fault tolerance and availability.