Ready to dive into the lake?
lakeFS is currently only
available on desktop.

For an optimal experience, provide your email below and one of our lifeguards will send you a link to start swimming in the lake!

lakeFS Community
Idan Novogroder
Idan Novogroder Author

Idan has an extensive background in software and DevOps engineering....

Published on July 9, 2024

Continuous integration/continuous delivery (CI/CD) helps software developers adhere to security and consistency standards in their code while meeting business requirements. Today, CI/CD is also one of the data engineering best practices teams use to keep their data pipelines efficient in delivering high-quality data.

What is a CI/CD pipeline and how do you implement it? Keep reading to find out. 

What are CI, CD, and continuous deployment?

Before we move on to explore CI/CD pipelines, let’s take a closer look at the approaches behind CI/CD.

What is continuous integration (CI)?

Using continuous integration (CI), teams can maintain version control while submitting numerous changes to a primary code branch or shared repository. It allows for rapid development while preventing merge conflicts, problems, and duplication. 

In addition to ensuring that the main branch is constantly up to date, database continuous integration can create temporary, isolated side or feature branches for minor database changes that may later be integrated into the main branch.

What is continuous delivery (CD)?

Continuous delivery is the process of automatically preparing code changes for release to production. It builds on continuous integration by deploying all code changes to a production environment following the build phase. 

Teams can validate application changes across several dimensions using continuous delivery, which enables developers to automate testing beyond unit tests. These tests can consist of load testing, integration testing, user interface testing, API reliability testing – all helping find problems early on and validate updates.

What is continuous deployment?

Code updates for an application are automatically deployed into the production environment as part of the continuous deployment process.

A set of pre-written tests serves as the engine for the automation. The system delivers fresh updates straight to users of the software after they pass those tests.

By removing the delay that usually occurs over several days, weeks, or even months between coding and customer value, continuous deployment accelerates time to market.

What is a CI/CD Pipeline?

A continuous integration/continuous delivery (CI/CD) pipeline is an automated code development and testing procedure combined with deployment. It allows teams to create, test, and deploy applications faster.

The process ensures that an application is always available to customers by automatically releasing it to its appropriate environment. Automating tests and builds guarantees that errors are found early and quickly corrected, preserving high-quality software or data.

Database schema and logic changes are integrated into your app development process using a CI/CD data pipeline. 

Understanding CI/CD as Write-Audit-Publish (WAP)

In data engineering, the Write-Audit-Publish (WAP) paradigm allows teams to have more control over the quality of the data. 

Ensuring data reliability for users is the fundamental goal of WAP. This is accomplished by verifying the data after it has been processed but before customers can access it. We refer to it as a pattern because the actual implementation will differ greatly depending on the particulars of your technology platforms, architectural layouts, and other requirements that can conflict.

Write-Audit-Publish is helpful because it ensures that users of the data can trust the information they use. Programmatically enforcing several aspects of data quality – such as whether NULLs are present where they shouldn’t be – is simple. Do fields fall inside the predicted ranges? By doing this, we prevent the erosion of confidence that might happen when processed data is made public and then removed or corrected once mistakes are found.

Here’s an example of how it works

Our users initially view only the data that is currently available. It’s waiting to process new data.

New data for 18 May 2023 is available. Data for previous day is loaded in the database and shown in downstream use.

There’s new data for May 18, 2023. The previous day’s data is imported into the database and displayed for further use.

Currently, Write-Audit-Publish is used to release the data in a regulated manner. In essence, the pattern’s name sums it up as follows:


Put the data you’re processing in a format that will not be read by customers later on. This could be a branch, a staging location, or something else entirely.

CI/CD as Write-Audit-Publish (WAP)


Verify the data through an audit to ensure that the requirements for data quality are met. Only the current data is visible to users of the published data.

CI/CD as Write-Audit-Publish (WAP)


Data is published by publishing it to the location where downstream users will read it.

CI/CD as Write-Audit-Publish (WAP)

Benefits of the Write-Audit-Publish (WAP) Approach in CI/CD

Using WAP design patterns in data pipelines has a number of important advantages.

Advantage How it Works
Improved Data Integrity and Quality To ensure that only verified and trustworthy data sources are released into production, the audit step carefully examines the data for correctness, completeness, and compliance with established standards. It also fixes any inconsistencies or abnormalities.
Enhanced Data Security WAP’s structured approach isolates audited and raw data, preventing sensitive and unvalidated data from being exposed too soon. This is important for upholding security requirements and data compliance across a range of businesses.
Enhanced Reliability Better error management and recovery procedures are possible by the division of the writing, auditing, and publishing phases. This clear division makes it easier to find and fix problems at every turn, which supports the data pipeline’s overall stability and dependability.
Operational Flexibility and Scalability WAP patterns’ modular architecture offers the freedom to change, upgrade, or scale individual components on their own. This enables the system to adjust to changing requirements without jeopardizing the stability of the pipeline as a whole.

Types of CI/CD Pipelines

Cloud-Native CI/CD Pipelines

Cloud-native computing makes use of contemporary cloud services, such as multicloud, serverless, and container orchestration, to mention a few. Apps that are native to the cloud are designed to operate there.

Teams may deploy to various settings because of the cooperation of these two ideas: cloud native and continuous integration. By integrating and testing code changes in shared source code repositories, cloud-native continuous integration (CI) supports contemporary development teams by fusing the concepts of continuous integration (CI) with cloud services. 

This method improves the software development lifecycle and guarantees effective software delivery by utilizing technologies like Kubernetes and container registries. Cloud-native continuous integration provides the automation teams require for speed and stability, and it is made to accommodate the cloud services and architectures that cloud-native teams employ.

Kubernetes-Native Pipelines

The goals of the CI/CD processes and the Kubernetes platform are to increase the development pace, automate tasks, and produce better software. 

Some essential elements of a CI/CD pipeline built with Kubernetes include:

  • Encapsulating application components and facilitating smooth integration through runtimes via containers like Docker.
  • When the CI/CD tool authorizes the containers, operating clusters will launch the containers for your software build.
  • Configuration management detects any recently introduced changes to the system and keeps track of all infrastructure setup details.
  • Code updates are maintained in a unified source code repository called a version control system (VCS). When a new change is pushed into a CI/CD tool’s repository, this creates the trigger that causes the pipeline to start.
  • By guaranteeing that the pipelines are free from potential security threats, security testing and audits preserve the balance between application security and quick development.
  • Thanks to full life cycle visibility, continuous monitoring and observability give developers access to relevant insights and metrics.

CI/CD Pipeline for a Monorepo

Using a monorepo, or “monolithic repository,” entails keeping all of your code for numerous projects in a single repository. It’s crucial to take into account the situations in which monorepos can cause conflict, particularly when developing, putting into practice, and managing continuous integration and delivery (CI/CD).

Since many projects, libraries, or services are entwined within the same repository, dependency management in a monorepo is essential. Proper management of dependencies will provide optimal performance and functionality of builds, as well as minimize difficulties for developers working on the project.

To check for dependency problems, including out-of-date libraries or version conflicts, you can combine a variety of tools with your CI and continuous delivery pipelines. 

Because storing several projects or components in a single repository has special requirements and obstacles, managing access control with a monorepo is another key point. Consider separating project components into a different repository and limiting access to them if there are any areas you don’t feel comfortable with the whole team seeing.

6 Steps to Building a CI/CD Pipeline – With Examples

1. Plan Your Environment

Having the right hardware and cloud infrastructure on hand is critical to the seamless operation of your CI/CD pipeline. These resources include virtual machines, servers, containers, and version control systems, all of which are essential to the successful and efficient execution of your build and deployment processes.

Using a trustworthy CI/CD platform, such Travis CI, GitHub Actions, or Jenkins, is essential to optimizing and automating the pipeline. Incorporating an artifact repository, like Nexus or JFrog Aritfactory, is crucial for efficiently managing and storing build artifacts and dependencies.

Finally, it is strongly advised to use Infrastructure as Code (IaC) tools such as Terraform or AWS CloudFormation.

Once the necessary tools have been determined, it is critical to configure your Version Control System (VCS) correctly to guarantee a seamless workflow. Create a repository on a VCS platform for your project first. Include the source code for your project in the initial commit to this repository. Moreover, you must design a branching strategy that is appropriate and compatible with your development process.

2. Define Your Workflow

If you want to create a pipeline with a systematic process for code changes, you need a well-defined CI/CD methodology. This coordinated approach ensures accuracy, reduces inconsistencies, and promotes productive teamwork. The software delivery process should be divided into doable, automated processes to properly design your pipeline stages.

Some of the major steps in your workflow may be:

  • Code compilation
  • Unit testing
  • Integration testing
  • Code analysis
  • Artifact creation
  • Deployment to staging environments
  • Automated testing
  • Quality assurance
  • Manual testing
  • Deployment to production 

Approval gates and rollback mechanisms are useful tools for managing possible problems that might occur during the pipeline.

3. Code Integration and Automated Testing

The code integration process is the cornerstone of any DevOps CI/CD pipeline. This crucial phase highlights the continuous and smooth integration of code changes into the project’s main codebase.

The deployment of a version control system, like Git, is essential to efficiently monitor code changes, enable smooth team member discussion, and create a central repository for your project’s codebase.

Essential scripts known as “Git hooks” run automatically either before or after particular Git events, like pushes or commits. Enforcing coding standards, running tests, and carrying out several other tasks automatically are all made possible by these hooks, which ensure that code uniformity and quality are maintained.

To keep the testing process consistent and effective, automation is essential. Tests can be carried out automatically each time changes are made to the code by using a variety of test automation frameworks and tools. Development and testing go hand in hand so that developers can get feedback quickly and make progress quickly.

4. Set Up Your CI Server

The central component in charge of coordinating the DevOps CI/CD pipeline is the Continuous Integration Server. The process of assembling code, running tests, and producing artifacts all depend on build jobs. CI servers supervise the smooth operation of these jobs and efficiently handle them in reaction to changes introduced to the codebase.

CI/CD systems automate operations that are triggered by changes to the code more efficiently. The process of configuration involves creating build jobs, deciding which tests to run, and putting triggers in place inside the pipeline.

5. Deployment Automation

Deployment automation ensures a seamless transition for your application from development to the final deployment stage.

Make sure to prepare the infrastructure and settings appropriately to guarantee a smooth deployment across many environments. This means that the components and settings required for development, staging, and production environments need to be set up. 

Automation scripts, which are often created with well-known tools like Ansible, Puppet, or Kubernetes YAML files, make deployment easier by guaranteeing reliable and repeatable deployments in a variety of scenarios.

6. Continuous Monitoring

It’s key that you use continuous monitoring to guarantee the best possible functioning of deployed applications in a production setting.

Configuring monitoring tools (like Grafana and Prometheus) is the first step. They provide real-time insight into the functionality of applications, the state of infrastructure, and the user experience.

Teams can gain from the smooth integration of automated warnings and responsive actions to handle production issues by integrating monitoring into the CI/CD pipeline. The proactive approach improves the system’s overall reliability and makes it easier to solve problems promptly.

Attributes of an Efficient CI/CD Pipeline


Every cycle uses the exact identical components and procedures that are employed in one. When a build produces a Docker container, for instance, that container is the item that is tested and sent through the pipeline for deployment or delivery. 

Developers can write scripts and set up automated procedures knowing that their efforts will be fruitful and consistently produce the required outcomes. Variations or manual stages in the process slow down the flow and increase the risk of mistakes and inefficiency


By facilitating quicker bug fixes, guaranteeing product stability, providing better operational support, lowering deployment failures, and reducing time to market, CI/CD pipelines help to increase the reliability of data pipelines. When taken as a whole, these elements provide a data processing environment that is more reliable and effective.


A build should be able to move quickly through the pipeline (integration, testing, delivery, and even deployment); it may only take a few hours to complete a testing cycle and a few minutes to complete an integration. 

Benefits of a CI/CD Pipeline

Using CI/CD pipelines has numerous advantages, such as:

  • Lower Costs – Fewer human resources are required when the software or infrastructure is developed and deployed faster.
  • Reduced Deployment Time – Automation shortens the deployment time. From coding to deployment, the entire process is optimized to shorten the overall process duration and increase its efficiency. 
  • Enabling Constant Feedback – CI/CD allows teams to make necessary improvements to their workflow and code in response to input. The testing phase finds errors or flaws and provides fast feedback so that the code can be fixed. Errors are easier and faster to rectify when they are found early in the development process. Team members can be made aware of wins and failures at every level by setting up notifications.
  • Enhanced Teamwork – The group may see comments, see issues as they arise, and adjust as necessary.
  • Audit Trails – Every step of the pipeline produces records, which can be used to provide traceability and accountability.

Challenges of a CI/CD pipeline

Business executives and development teams nevertheless need to take into account some of the possible drawbacks of CI/CD pipelines, despite their strong advantages:

  • Commitment to Automation – CI/CD depends on robust automation to create, test, and deploy every build, as well as consistency in an established toolset. The implementation and management of automation can require a significant commitment and may include a challenging learning curve.
  • Planning and Discipline –  If CI/CD isn’t continuously producing new builds, testing release candidates, and putting chosen candidates into production, it can’t provide the business with the full benefits it promises. Careful preparation and proficient project management abilities are needed for this. Following approved development rules is necessary for developers to guarantee standards of quality, style, and architecture.
  • Cooperation and Dialogue –  Effective communication and collaboration between developers and project stakeholders are crucial, regardless of the level of automation and tooling employed.

CI/CD Pipeline Best Practices

Single Source Repository

Consolidate all of your documentation, configuration files, and code into a single version control system. This facilitates team collaboration, tracking of changes, and management.

Build Once

You run the danger of introducing inconsistencies when you rebuild the code for multiple environments, and you can’t be sure that all of the prior tests succeeded. Rather, every build pipeline step should promote the same build artifact, which should then be made live.

To implement this, the build must be independent of any particular system. Rather than being included in the build process, any variables, authentication parameters, configuration files, or scripts should be called by the deployment script. This makes it possible to test the same build in several settings with the same build, boosting team confidence in that specific build artifact at each stage.

Prioritize Automation Efforts

In your CI/CD pipeline, try to automate as many processes as you can, from building and testing to deployment and monitoring. This will save time, increase uniformity, and lessen the possibility of human error.

Test Early and Often

By providing smaller updates more frequently, CI aims to simplify the process of merging changes from multiple contributors.

A series of automated tests are triggered by every commit to give quick feedback on the modification. Regularly committing reduces the chance of unpleasant merge conflicts when integrating significant, complicated changes and guarantees that everyone on your team is operating from the same principles, which promotes teamwork.

To fully benefit from CI, all team members must push their changes to main (master) so that others can see them, and they must update their working copy to receive updates from others. Try to commit to main (master) at least once a day as a general guideline.

Make the CI/CD Pipeline the Only Way to Deploy

Usually, the request to bypass the release procedure is made when there is a little or urgent modification (or both), but caving in to these demands can become troublesome. 

For example, you might introduce avoidable flaws by skipping the automated quality assurance stages. It’s more difficult to reproduce and debug faults because the build is not easily available for deployment to a testing environment.

Demand Visibility

Your CI/CD tool’s metrics analysis allows you to pinpoint possible problems and areas in need of development.

You need monitoring to learn how your pipeline infrastructure is being utilized, when peak load tends to occur, and whether you need to scale it up or down by comparing the number of builds that are triggered every week, day, or hour.

You may find opportunities to streamline your quality assurance coverage by going through your QA results and identifying the ones that are frequently disregarded.

Clean Environments with Every Release

It gets more difficult to maintain track of all the configuration updates and modifications that have been made to each environment when they are run for extended periods.

Tests that pass or fail in one may not provide the same result in another as parameters diverge over time from both the original setup and from one another. The expense of maintaining static environments must be considered to avoid slowing down the quality assurance and release processes.

CI/CD Pipeline Solutions


Bamboo by Atlassian automates the management of software application releases. Build and functional testing, versioning, tagging releases, deploying, and activating new versions on production are all covered. It detects and automatically applies the main line’s CI scheme to newly created branches in SVN, Mercurial, and Git repos.


CircleCI facilitates quick software development and publication. Automation is possible throughout the user’s pipeline using CircleCI, from code development and testing to deployment.

When new code lines are committed, builds can be automatically generated by CircleCI through integration with Bitbucket, GitHub, and GitHub Enterprise. In addition, CircleCI offers cloud-managed continuous integration services and operates on private infrastructure behind a firewall.


A collection of technologies called GitLab is used to manage several facets of the software development lifecycle. The main offering is a web-based Git repository manager that comes equipped with Wiki, analytics, and issue-tracking capabilities.

With every contribution or push in GitLab, you may initiate builds, execute tests, and make application code available. Jobs can be built on a different server, in a Docker container, or a virtual machine.

Microsoft Azure DevOps

Azure DevOps is a Software-as-a-Service platform with an entire toolkit that supports not only DevOps functions but also the skills needed to manage the full product development lifecycle.

Azure DevOps’ versatility is one of its best features since it can be integrated with other products on the market to manage process flow jointly and coordinate the entire DevOps toolchain. Amazon Web Services (AWS) and Google Cloud Platform (GCP) are also supported by Azure DevOps’ Continuous Integration and Delivery architecture.

CI/CD Tools

Continuous Integration Tools

  • Jenkins – A popular open-source continuous integration tool that builds, integrates, and tests the code automatically. Jenkins has the Docker plug-in available.
  • Buildbot – This software development tool is capable of automating every step of the process. It queues, completes, and reports jobs as a job scheduling system.
  • Travis CI – One of the most reputable and established hosted solutions, it is also offered in an enterprise on-premises version.
  • GitLab CI – A free hosted service that is an essential component of the open-source Rails project. This program offers comprehensive Git repository management along with other capabilities like code reviews, issue tracking, and access control.

Continuous Delivery and Deployment Tools

  • DevOps on Microsoft Azure – Through the combination of Azure’s power and DevOps’ flexibility, teams can design, manage, and deploy apps quickly and efficiently. With its extensive toolkit for automation, continuous integration, and delivery, teams can create and launch apps more rapidly. 
  • Build with Google Cloud – It automates software creation, testing, and deployment, and developers can more easily and quickly push code changes to production. Along with integrating with other GCP services like App Engine, Cloud Storage, BigQuery, and Cloud Spanner, it supports several languages, development environments, and runtimes.
  • CodeDeploy on AWS – You can use it to automate software deployments to Amazon EC2 instances, on-premises instances, serverless Lambda functions, or Amazon ECS services can be automated with the help of the AWS CodeDeploy service. 
  • CircleCI – A collection of continuous integration and delivery (CI/CD) technologies intended to facilitate the efficient development, testing, and deployment of software. It offers an all-inclusive platform that streamlines the software development, testing, and deployment processes, freeing up developers to concentrate on producing the greatest potential end product.

Machine Learning CI/CD Applications

  • CML – It aims to accelerate the delivery of ML models with fewer defects and to simplify their deployment and implementation. The goal of CML is to automate some of the repetitive operations that machine learning engineers have to complete daily. These jobs include training models, assessing their performance, building and labeling datasets, and more. 
  • GitHub Actions – Its CI/CD functionalities include code pushes, releases, and problem management. Its CI/CD capabilities include automated testing, container development, web service deployment, and automating new user onboarding for your open-source project.
  • TeamCity – A continuous integration server that offers a fairly robust free edition for modest projects (up to 100 build configurations) and a wide range of open-source plugins, including those from JetBrains and other developers, are fully supported by TeamCity.

CI/CD in the Cloud


To speed up software development and release cycles, AWS provides a full suite of CI/CD developer tools. Based on the specified release model, AWS CodePipeline automates the build, test, and deploy stages of the release process for each code update. This makes it possible to deploy features and updates consistently and quickly.

Code pipelines are capable of service integration. These could be third-party goods like GitHub or AWS Services like Amazon Simple Storage Service (Amazon S3). 

CI/CD in Azure

Azure DevOps is a comprehensive toolbox to support both the skills required to manage the whole product development lifecycle and DevOps functions. One of Azure DevOps’ main characteristics is its versatility; it can be linked with other solutions on the market to coordinate the whole DevOps toolchain and manage process flow collaboratively. 

CI/CD in Azure

CI/CD in Google Cloud

Cloud Build is a fully managed continuous integration and delivery platform that lets you develop, test, and deploy applications. Cloud Build can import source code from Bitbucket, GitHub, Cloud Storage, and Cloud Source Repositories and generate artifacts like Docker container images or Java archives.

Your build is carried out by Cloud Build as a sequence of build steps, each of which is carried out in a Docker container. Any task that can be completed from a container, regardless of the environment, can be completed by a construction step.

CI/CD in Google Cloud

CI/CD in IBM Cloud

You can automate app development and deployment by setting up continuous integration and delivery (CI/CD), version control, tool chains, and more with IBM Cloud and other open-source tools.

IBM Continuous Delivery Pipeline for IBM Cloud helps teams build a solid DevOps methodology. This platform offers open toolchains that automate the development and deployment of containerized applications.

CI/CD Pipeline KPIs

Development Frequency

The frequency with which you use your CI/CD pipeline to deploy to production is tracked by deployment frequency. A high deployment frequency suggests fewer modifications per deployment. Low deployment frequency may indicate that commits are not being put into the pipeline frequently, possibly due to jobs not being divided into smaller units, or it may indicate that updates are being bundled into larger releases.

Change Lead Time

Change lead time is the duration between a feature’s initial proposal and its user release. Lead time is typically extremely flexible when it comes to ideation, user research, and prototyping. A lengthy lead time prevents you from constantly putting code changes in front of consumers and, consequently, from obtaining input to improve what you’re creating. 

Change Failure Rate

The percentage of changes that are put into production and fail is known as the change failure rate. This metric’s benefit is that it contextualizes failed deployments in relation to the number of modifications made.


MTTR (mean time to recovery or resolution) refers to the amount of time needed to fix a production failure. Maintaining a low MTTR call for proactive monitoring of your system in production to notify you of issues as soon as they arise and the capacity to quickly roll back modifications or implement a remedy via the pipeline.

Mean time to detection (MTTD), a similar metric, calculates the interval of time between the deployment of a modification and the discovery of a problem by your monitoring system as a result of that change. 

CI/CD Pipelines and lakeFS

lakeFS offers a data version control system analogous to Git, which simplifies the setup of CI CD pipelines for data. Managing CI/CD pipelines with lakeFS involves several key practices to streamline data workflows. lakeFS offers hooks that automate data validations and checks at critical points, such as during commits and merges, ensuring data integrity and consistency. These hooks are similar to Git Hooks but are specifically designed to handle data operations in a scalable and reliable manner.

lakeFS allows users to create isolated environments for development and testing, ensuring that changes can be thoroughly evaluated before being merged into production. This approach helps prevent data corruption and ensures that only validated and tested data reaches the production environment.

By integrating lakeFS with existing CI/CD tools, users can leverage its version control capabilities to manage data changes efficiently. The system’s ability to track data versions and provide reproducible environments is crucial for maintaining the accuracy and reliability of data workflows.

Overall, lakeFS enhances CI/CD pipelines by providing robust data management features, ensuring automated and reliable data operations, and supporting seamless integration with other development tools, thereby improving the efficiency and effectiveness of data engineering processes.

CI/CD Data Pipeline with lakeFS

Learn more: CI/CD for Data Lakes | lakeFS Documentation 


Whether it’s a CI/CD pipeline for source code or database management, you can draw multiple benefits from these best practices and tooling. For the latter case, such a setup keeps everything flexible by automatically updating the database schema during the delivery or deployment process. 

Learn more about building CI/CD pipelines for data here: CI/CD for data pipelines – The Shortest Path to Your Destination with lakeFS.

Git for Data – lakeFS

  • Get Started
    Get Started
  • Did you know that lakeFS is an official Databricks Technology Partner? Learn more about -

    lakeFS for Databricks