Glossary

This glossary defines more than 90 terms related to project management, development, source control, data, and BI. Each entry provides a short definition (🟢) and links…

Guide≈ 13 min readPartly checkedReading

In brief

This glossary defines more than 90 terms related to project management, development, source control, data, and BI. Each entry provides a short definition (🟢) and links to the file that explores it in depth (🟡/🔴). Acronyms are listed under their usual form. When a term has variants (Scrum “master” or “facilitator,” Import mode or DirectQuery), the definition uses common usage.

On this page

A
B
C
D
E
F
G
H
I
J
K
L
M
N
O
P
Q
R
S
T
U
V
W
X
See also

Level: 🟢→🔴

A#

Agile — Family of iterative approaches based on the Agile Manifesto (2001): deliver early and often, adapt to change, collaborate with the customer. It is not a method but a mindset applied through Scrum, Kanban, XP… → Agile and Lean Methods (Manifesto, Kanban, XP, Lean, Shape Up, DSDM, SAFe, DevOps…)
ALM (Application Lifecycle Management) — Management of an application's lifecycle: requirements, planning, code, testing, delivery, operations, within an integrated toolchain. Azure DevOps is a typical example. → Azure DevOps Server
Artifact (Scrum) — Transparency object in the Scrum framework: Product Backlog, Sprint Backlog, and Increment, each tied to a commitment (Product Goal, Sprint Goal, Definition of Done). → Scrum, the framework (Scrum Guide 2020)
Azure DevOps Server — Successor to Team Foundation Server (TFS): the on-premises installable version of Azure DevOps (work items, Git/TFVC repositories, pipelines, tests, artifacts). → Azure DevOps Server (ex-TFS) in Depth
Partial achievement — Progress on a maturity milestone at 0, 0.5, or 1, which allows rating a level without requiring perfection. → RATING

B#

Backlog — Ordered list of remaining work (Product Backlog: the product; Sprint Backlog: the current sprint). The order reflects value and risk, not age. → Scrum in Practice
Big bang (delivery) — Go-live all at once, with no stages. Maximum risk, difficult rollback; opposite of incremental delivery. → Predictive Methods (waterfall, V-model, spiral, RUP, PMBOK, PRINCE2, Merise, CMMI)
BI (Business Intelligence) — Set of techniques that transform operational data into decision indicators: warehouse, semantic model, reports. → BI Development Matrix
Branch — Parallel development line in a version control system; in Git it is a simple lightweight pointer, in TFVC a logical copy of the folder. → Branching Strategies
Burndown / burnup — Charts of remaining work (burndown) or completed work (burnup) over time; burnup makes scope changes visible. → Project Metrics and Reporting
Build — Automated compilation and assembly step; for data, generation of a dacpac, an installation kit, or a report project. → CI/CD: Continuous Integration and Continuous Delivery

C#

Cadence — Regular rhythm of events or deliveries (two-week sprint, monthly release). → Scrum in Practice
Capacity (Fabric / Power BI) — Reserved compute resource (SKU F; P being phased out; A/EM) that runs models, notebooks, and pipelines; its sizing governs performance and cost. → Cost, Licensing, and Capacity (BI)
Waterfall — Sequential lifecycle: requirements, design, implementation, testing, delivery. Effective when needs are stable and the cost of error is high. → Predictive Methods (waterfall, V-model, spiral, RUP, PMBOK, PRINCE2, Merise, CMMI)
CD (Continuous Delivery / Deployment) — Continuous delivery: every validated change is deployable on demand (delivery) or automatically deployed (deployment). → CI/CD: Continuous Integration and Continuous Delivery
Changeset (TFVC) — Atomic batch of changes stored in TFVC, numbered in increasing order; approximate equivalent of a commit, but centralized. → TFVC in Depth
CI (Continuous Integration) — Frequent integration of everyone's work into a shared branch, validated by automated build and tests. → CI/CD: Continuous Integration and Continuous Delivery
CMMI — Process maturity model with levels (1 to 5), predecessor of maturity matrices; distinct from the matrices in this section, which measure outcomes. → Predictive Methods (waterfall, V-model, spiral, RUP, PMBOK, PRINCE2, Merise, CMMI)
Commit — Snapshot saved in Git, identified by a hash and a parent; the unit of history. → Git In Depth
Data contract — Explicit agreement between producer and consumer on the schema, semantics, freshness, and quality of a dataset. → Data Sources and Contracts
Change management — Human support for a change (training, communication, sponsors); a condition for report adoption. → Adoption and Change Management
CRISP-DM — Data mining lifecycle in six phases (business understanding, data understanding, preparation, modeling, evaluation, deployment), version 1.0 from 2000. → Methods Specific to Data Projects
V-model — Waterfall variant where each design phase has a symmetric testing phase. → Predictive Methods (waterfall, V-model, spiral, RUP, PMBOK, PRINCE2, Merise, CMMI)
Cycle time — Duration between the actual start of work on an item and its completion; not to be confused with lead time. → Project Metrics and Reporting

D#

dacpac — SQL Server data-layer package (database schema) used to compare and deploy declaratively. → CI/CD for data: foundations, datamarts, pipelines, and reports
DAX (Data Analysis Expressions) — Formula language for the tabular model (Power BI, Analysis Services): measures, calculated columns, calculated tables. → DAX Measures and Calculations
Data lineage — See Lineage.
Data warehouse / warehouse — Integrated, historized database organized by subject, dedicated to analysis. → Modeling and Architecture
Datamart — Thematic subset of a warehouse, often modeled as a star schema, serving a business domain (sales, readings). → Modeling and Architecture
Dataflow Gen2 — Low-code Power Query transformation in Fabric that writes to a lakehouse or a warehouse. → OneLake — architecture
DataOps — Application of DevOps practices (versioning, testing, CI/CD, monitoring) to data pipelines. → Lead Time and Industrialization (DataOps)
Definition of Done (DoD) — Shared criteria that make an item “done” (tested, reviewed, documented, deployable); commitment of the Increment. → Scrum, the framework (Scrum Guide 2020)
Definition of Ready (DoR) — Conditions for an item to enter a sprint (clear, estimated, testable); useful practice but absent from the Scrum Guide. → Scrum in Practice
Technical debt — Future cost caused by present shortcuts (duplicated code, missing tests, outdated documentation). → Technical Debt and Maintainability
DevOps — Culture and practices that bring development and operations closer to deliver fast and reliably: automation, fast feedback, shared responsibility. → Agile and Lean Methods (Manifesto, Kanban, XP, Lean, Shape Up, DSDM, SAFe, DevOps…)
Direct Lake — Power BI/Fabric mode that reads Delta files directly from the lakehouse in OneLake, without import; the SQL endpoint variant can fall back to DirectQuery. → OneLake — architecture
DirectQuery — Mode where each visual queries the source live: maximum freshness, performance dependent on the source. → Comparison — OneLake / Fabric vs Managed SQL Server for Power BI Reporting
DORA (metrics) — Four delivery performance measures from DORA research (book Accelerate): deployment frequency, lead time for production, change failure rate, time to restore; the DORA 2024 model adds deployment rework rate and refers to time to restore after a failed deployment. → Project Metrics and Reporting
DSDM — Agile method with variable scope and fixed time (time-boxing). → Agile and Lean Methods (Manifesto, Kanban, XP, Lean, Shape Up, DSDM, SAFe, DevOps…)

E#

ELT / ETL — Extract, transform, load (ETL) or extract, load, transform in the target (ELT). The latter dominates with cloud engines. → Data Layer and ETL
Epic — Large body of work broken down into user stories across multiple sprints. → Scrum in Practice
Estimation (story points) — Relative effort assessment, not in hours; used for forecasting, not for evaluating people. → Planning and Estimation (PLA)
Star (schema) — Dimensional model: one fact table surrounded by dimensions; foundation of any solid semantic model. → Semantic Model
Eventstream — Fabric feature for capturing and routing real-time event streams. → OneLake — architecture

F#

Fabric (Microsoft Fabric) — Microsoft's unified SaaS analytics platform (OneLake, lakehouse, warehouse, pipelines, Power BI, real time). Evolving offering (status 2026-10): check GA/preview status for each feature. → OneLake + SaaS Mobile Apps: Overview
Feature flag — Switch that enables or disables a feature without redeploying; decouples deployment and release. → CI/CD: Continuous Integration and Continuous Delivery
Merge (merge) — Integration of changes from one branch into another; the difficulty lies in conflicts, especially in model files. → Git In Depth
Flow (flow, Kanban) — Continuous movement of work; managed through work-in-progress limits and throughput. → Agile and Lean Methods (Manifesto, Kanban, XP, Lean, Shape Up, DSDM, SAFe, DevOps…)

G#

VCS (version control system) — See SCM. A system that keeps file history and coordinates parallel work (Git, TFVC). → Overview of Version Control Systems
Git — Distributed VCS: each clone contains the full history; lightweight branches, frequent merges. → Git In Depth
Data governance — Roles, rules, and processes that ensure data ownership, quality, security, and compliant use. → Data Governance and Organization
Gated check-in — TFVC control where check-in is accepted only if a build succeeds. → TFVC in Depth

H#

Hybrid (method) — Deliberate combination of predictive and agile elements (waterfall framing, iterative delivery). → Choosing Your Method and Hybridizing
Hotfix — Urgent fix applied to a production version, then carried over to the development branch. → Branching Strategies

I#

Generative AI / LLM — Language models used as assistants (code, SQL, DAX, documentation, review). Real gains, real limits: hallucinations, privacy. → Cross-Functional Summary — AI Contribution
Idempotence — Property of a process that, when rerun, produces the same result; foundation of safe load replay. → CI/CD for data: foundations, datamarts, pipelines, and reports
Import (mode) — Mode where data is copied into the in-memory model: fast in use, refresh must be scheduled. → Comparison — OneLake / Fabric vs Managed SQL Server for Power BI Reporting
Increment — Concrete, usable outcome, accumulated at the end of a sprint. → Scrum, the framework (Scrum Guide 2020)
Iteration — Short work loop that produces a verifiable outcome (sprint, timebox). → Project Management Fundamentals

J#

Milestone (maturity) — Precise, verifiable criterion for a matrix level, with its stable identifier (e.g., PI-RSK-3.2). → RATING
Jira — Atlassian ticket and project tracking tool, highly configurable. → Jira In Depth and Alternatives

K#

Kanban — Flow management method: visualize work, limit work in progress, measure, improve. → Agile and Lean Methods (Manifesto, Kanban, XP, Lean, Shape Up, DSDM, SAFe, DevOps…)
Kimball (method) — Dimensional modeling approach (The Data Warehouse Toolkit): business processes, grain, dimensions, facts. → Modeling and Architecture
Installation package — Versioned, replayable set of scripts that installs or updates a database for a client. → Source Control for Data: SQL and Power BI
KPI — Key performance indicator; useful only with a definition, a source, and an owner. → Project Metrics and Reporting

L#

Lakehouse — Architecture that combines the flexibility of a file lake with the transactional tables of a warehouse (Delta tables). → OneLake + SaaS mobile apps: architecture
Lead time — Total time between request and delivery; measures what the customer perceives. → Project Metrics and Reporting
Lean — Philosophy from industry: eliminate waste, decide at the last responsible moment, deliver fast. → Agile and Lean Methods (Manifesto, Kanban, XP, Lean, Shape Up, DSDM, SAFe, DevOps…)
Lineage — Traceability of data from source to visual: where it comes from, what it went through. → Metadata, Lineage, and Documentation
Lineage tag — Internal identifier of a Power BI model object; changing it breaks the link with reports. → Versioning and Industrialization of Reports

M#

Maturity matrix — Themes × five levels grid; each level is defined by a promise and verifiable milestones. → Maturity Matrices — Principle, Anatomy, and Comparison
Medallion (bronze/silver/gold) — Layered lakehouse organization: raw, cleaned, analysis-ready. → OneLake + SaaS mobile apps: architecture
Merise — French information systems design method (MCD, MLD, MPD). → Predictive Methods (waterfall, V-model, spiral, RUP, PMBOK, PRINCE2, Merise, CMMI)
Metadata — Data about data: definitions, owners, types, lineage. → Metadata, Lineage, and Documentation
Mirroring (Fabric) — Continuous replication of an operational database to OneLake, without writing a pipeline. → SQL Server in SaaS — architecture
Semantic model — Business layer of a Power BI report: tables, relationships, hierarchies, measures, security. → Semantic Model
MTTR — Mean time to restore after an incident; resilience metric. → Observability and Reliability
MVP (minimum viable product) — Smallest version that allows learning from the market; not to be confused with a throwaway prototype. → Project Management Fundamentals

N#

Maturity level — From 1 to 5: each level describes an observable outcome, not a means. → RATING
NFR (non-functional requirements) — Performance, security, availability, accessibility: often forgotten during framing. → Scope and Requirements Management (PER)

O#

OneLake — Fabric’s single logical data lake, in open Delta/Parquet format, “OneDrive for data.” → OneLake + SaaS Mobile Apps: Overview
OKR — Objectives and key results: framework for aligning priorities. → Governance and Steering (GOU)

P#

PaaS / SaaS — Managed platform (Azure SQL Database: you manage data, not the server) versus complete software (Power BI service, Fabric). → SQL Server in SaaS — overview
PBIP / TMDL — Power BI project format in text files (PBIP) and model description language (TMDL): versionable and mergeable. → Versioning and Industrialization of Reports
Pipeline (data) — Orchestrated chain that moves and transforms data. → CI/CD for data: foundations, datamarts, pipelines, and reports
PMBOK — PMI project management guide: processes, performance domains. → Predictive Methods (waterfall, V-model, spiral, RUP, PMBOK, PRINCE2, Merise, CMMI)
Power BI — Microsoft BI suite: Desktop (authoring), service (publishing), semantic model, reports. → Cross-Cutting Power BI Good Practices
PRINCE2 — Project management method structured by stages, roles, and continuous business justification. → Predictive Methods (waterfall, V-model, spiral, RUP, PMBOK, PRINCE2, Merise, CMMI)
Product Owner — Accountable for product value and Product Backlog ordering. → Scrum, the framework (Scrum Guide 2020)
Pull request — Request to merge a Git branch with code review. → Git In Depth

Q#

Data quality — Fitness of data for use: accuracy, completeness, consistency, freshness, uniqueness, validity. → Data Quality
Quality gate — Automatic threshold to pass (tests, coverage, static analysis) before proceeding. → CI/CD: Continuous Integration and Continuous Delivery

R#

Rebase — Rewriting a Git branch onto another base to linearize history. → Git In Depth
Acceptance testing — Delivery validation by the business: for data, reconciliation of figures against a reference. → Acceptance testing and business validation
Refactoring — Improving code structure without changing behavior. → Code Quality
Retrospective — Scrum event for improving how the team works. → Scrum in Practice
GDPR — European regulation for personal data protection: minimization, legal basis, data subject rights. → Security, Privacy, Compliance
RLS (Row-Level Security) — Row-level security: a user sees only rows in their scope (region, customer). → RLS Security and Access
RUP — Rational Unified Process: iterative method driven by use cases. → Predictive Methods (waterfall, V-model, spiral, RUP, PMBOK, PRINCE2, Merise, CMMI)

S#

SAFe — Agile scaling framework (trains, PI Planning). → Agile and Lean Methods (Manifesto, Kanban, XP, Lean, Shape Up, DSDM, SAFe, DevOps…)
SCD (Slowly Changing Dimension) — Dimension historization technique (type 1: overwrite; type 2: version). → Modeling and Architecture
SCM — Source Code Management, synonymous with VCS. → Overview of Version Control Systems
Scrum — Sprint-based agile framework, three accountabilities (Product Owner, Scrum Master, Developers), five events, three artifacts. → Scrum, the framework (Scrum Guide 2020)
Scrum Master — Ensures the effectiveness of the framework and the team, not a project manager. → Scrum in Practice
SFA (Sales Force Automation) — Sales force automation application, a typical source for this section's common thread. → OneLake + SaaS Mobile Apps: Overview
Shape Up — Six-week cycle method with betting and “appetite.” → Agile and Lean Methods (Manifesto, Kanban, XP, Lean, Shape Up, DSDM, SAFe, DevOps…)
Shelveset (TFVC) — Set-aside of uncommitted changes, shareable for review. → TFVC in Depth
Shortcut (OneLake) — Shortcut that exposes external data in OneLake without copying it. → OneLake + SaaS mobile apps: architecture
Sprint — Fixed-length period (one month max) that produces an increment. → Scrum, the framework (Scrum Guide 2020)
SSDT — SQL Server Data Tools: versionable database projects, dacpac generation. → Source Control for Data: SQL and Power BI
Branching strategy — Convention for branch usage (trunk-based, GitFlow, release branches). → Branching Strategies

T#

TDD — Test-Driven Development: write the test before the code. → Testing
TFVC — Team Foundation Version Control: Microsoft’s centralized VCS, optional locks, changeset-based history. → TFVC in Depth
Trunk-based development — Daily integration into a main branch, very short-lived branches. → Branching Strategies

U#

User story — Need expressed from the user’s point of view, with acceptance criteria. → Scrum in Practice

V#

Velocity — Quantity of items (points) completed per sprint; a team forecasting tool, never for cross-team comparison. → Project Metrics and Reporting
Lock-in — Dependency on a vendor that increases exit costs; an architecture selection criterion. → Comparison — OneLake / Fabric vs Managed SQL Server for Power BI Reporting

W#

WIP (Work In Progress) — Work in progress: started but unfinished items; limiting it reduces lead times. → Agile and Lean Methods (Manifesto, Kanban, XP, Lean, Shape Up, DSDM, SAFe, DevOps…)
Work item — Azure DevOps work item (task, bug, user story). → Azure DevOps Server (ex-TFS) in Depth

X#

XP (Extreme Programming) — Agile engineering practices: TDD, pair programming, continuous integration, refactoring. → Agile and Lean Methods (Manifesto, Kanban, XP, Lean, Shape Up, DSDM, SAFe, DevOps…)

Glossary

A#

B#

C#

D#

E#

F#

G#

H#

I#

J#

K#

L#

M#

N#

O#

P#

Q#

R#

S#

T#

U#

V#

W#

X#

See also#