- AWS Implementation patterns information
- AWS known issues list
- Official GitLab releases as AMIs
- Additional details on implementation patterns
- Supplementary implementation patterns
- Intended audiences and contributors
GitLab Reference Architectures give qualified and tested guidance on the recommended ways GitLab can be configured to meet the performance requirements of various workloads. Reference Architectures are purpose-designed to be non-implementation specific so they can be extrapolated to as many environments as possible. This generally means they have a highly-granular “machine” to “server role” specification and focus on system elements that impact performance. This is what enables Reference Architectures to be adaptable to the broadest number of supported implementations.
Implementation patterns are built on the foundational information and testing done for Reference Architectures and allow architects and implementers at GitLab, GitLab Customers, and GitLab Partners to build out deployments with less experimentation and a higher degree of confidence that the results will perform as expected. A more thorough discussion of implementation patterns is below in Additional details on implementation patterns.
The following are the currently available implementation patterns for GitLab when it is implemented on AWS.
GitLab Site Reliability Engineering (SRE) for AWS - information for planning, implementing, upgrading and long term management of GitLab instances and runners on AWS.
Provision GitLab Cloud Native Hybrid on AWS EKS (HA). This document includes instructions, patterns, and automation for installing GitLab Cloud Native Hybrid on AWS EKS. It also includes Bill of Materials listings and links to Infrastructure as Code. GitLab Cloud Native Hybrid is the supported way to put as much of GitLab as possible into Kubernetes.
Omnibus GitLab on AWS EC2 (HA) - instructions for installing GitLab on EC2 instances. Manual instructions to build a GitLab instance or create your own Infrastructure as Code (IaC).
EKS Cluster Provisioning Patterns - considerations for setting up EKS cluster for runners and for integrating.
The following repository is self-contained in regard to enabling this pattern: GitLab HA Scaling Runner Vending Machine for AWS EC2 ASG. The feature list for this implementation pattern is good to review to understand the complete value it can deliver.
The Guided Explorations’ subgroup for AWS contains a variety of working example projects for:
- Using GitLab and AWS together.
- Running GitLab infrastructure on AWS.
- Retrieving temporary credentials for access to AWS services.
Known issues are gathered from within GitLab and from customer reported issues. Customers successfully implement GitLab with a variety of “as a Service” components that GitLab has not specifically been designed for, nor has ongoing testing for. While GitLab does take partner technologies very seriously, the highlighting of known issues here is a convenience for implementers and it does not imply that GitLab has targeted compatibility with, nor carries any type of guarantee of running on the partner technology where the issues occur. Please consult individual issues to understand GitLabs stance and plans on any given known issue.
See the GitLab AWS known issues list for a complete list.
GitLab produces AMI images during the regular release process. The AMIs can be used for single instance GitLab installation or, by configuring
/etc/gitlab/gitlab.rb, can be specialized for specific GitLab service roles (for example a Gitaly server). Older releases remain available and can be used to migrate an older GitLab server to AWS.
Currently the Amazon AMI uses the Amazon prepared Ubuntu AMI (x86 and ARM are available) as its starting point.
Instances running on Community Edition (CE) require a migration to Enterprise Edition (EE) in order to subscribe to the GitLab Premium or Ultimate plan. If you want to pursue a subscription, using the Free-forever plan of Enterprise Edition is the least disruptive method.
- Log in to the AWS Web Console, so that clicking the links in the following step take you directly to the AMI list.
Pick the edition you want:
- GitLab Enterprise Edition: If you want to unlock the enterprise features, a license is needed.
- GitLab Community Edition: The open source version of GitLab.
- GitLab Premium or Ultimate Marketplace (Prelicensed): 5 user license built into per-minute billing.
- AMI IDs are unique per region, so once you’ve loaded one of the above, select the desired target region in the upper right of the console to see the appropriate AMIs.
- After the console is loaded, you can add additional search criteria to narrow further. For instance, type
13.to find only 13.x versions.
- To launch an EC2 Machine with one of the listed AMIs, check the box at the start of the relevant row, and select the “Launch” button near the top of left of the page.
GitLab implementation patterns build upon GitLab Reference Architectures in the following ways.
Testing-backed architectural qualification is a fundamental concept behind implementation patterns:
- Implementation patterns maintain GitLab Reference Architecture compliance and provide GitLab Performance Tool (gpt) reports to demonstrate adherence to them.
- Implementation patterns may be qualified by and/or contributed to by the technology vendor. For instance, an implementation pattern for AWS may be officially reviewed by AWS.
- Implementation patterns may specify and test Cloud Platform PaaS services for suitability for GitLab. This testing can be coordinated and help qualify these technologies for Reference Architectures. For instance, qualifying compatibility with and availability of runtime versions of top level PaaS such as those for PostgreSQL and Redis.
- Implementation patterns can provided qualified testing for platform limitations, for example, ensuring Gitaly Cluster can work correctly on specific Cloud Platform availability zone latency and throughput characteristics or qualifying what levels of available platform partner local disk performance is workable for Gitaly server to operate with integrity.
Implementation patterns enable platform-specific terminology, best practice architecture, and platform-specific build manifests:
- Implementation patterns are more vendor specific. For instance, advising specific compute instances / VMs / nodes instead of vCPUs or other generalized measures.
- Implementation patterns are oriented to implementing good architecture for the vendor in view.
- Implementation patterns are written to an audience who is familiar with building on the infrastructure that the implementation pattern targets. For example, if the implementation pattern is for GCP, the specific terminology of GCP is used - including using the specific names for PaaS services.
- Implementation patterns can test and qualify if the versions of PaaS available are compatible with GitLab (for example, PostgreSQL, Redis, etc.).
Platform as a Service options are a huge portion of the value provided by Cloud Platforms as they simplify operational complexity and reduce the SRE and security skilling required to operate advanced, highly available technology services. Implementation patterns can be prequalified against the partner PaaS options.
- Implementation patterns help implementers understand what PaaS options are known to work and how to choose between PaaS solutions when a single platform has more than one PaaS option for the same GitLab role.
- For instance, where reference architectures do not have a specific recommendation on what technology is leveraged for GitLab outbound email services or what the sizing should be - a Reference Implementation may advise using a cloud providers Email as a Service (PaaS) and possibly even with specific settings.
Cost engineering is a fundamental aspect of Cloud Architecture and frequently the savings capabilities available on a platform exert strong influence on how to build out scaled computing.
- Implementation patterns may define GPT tested autoscaling for various aspects of GitLab infrastructure, including minimum idling configurations and scaling speeds.
- Implementation patterns may provide GPT testing for advised configurations that go beyond the scope of reference architectures, for instance GPT tested elastic scaling configurations for Cloud Native Hybrid that enable lower resourcing during periods of lower usage (for example on the weekend).
- Implementation patterns may engineer specifically for the savings models available on a platform provider. An AWS example would be maximizing the occurrence of a specific instance type for taking advantage of reserved instances.
- Implementation patterns may leverage ephemeral compute where appropriate and with appropriate customer guidelines. For instance, a Kubernetes node group dedicated to runners on ephemeral compute (with appropriate GitLab Runner tagging to indicate the compute type).
- Implementation patterns may include vendor specific cost calculators.
Implementation patterns are one step closer to specifics that can be used as a source for build instructions and automation code:
- Implementation patterns enable builders to generate a list of vendor specific resources required to implement GitLab for a given Reference Architecture.
- Implementation patterns enable builders to use manual instructions or to create automation to build out the reference implementation.
Implementation patterns may also provide specialized implementations beyond the scope of reference architecture compliance, especially where the cost of enablement can be more appropriately managed.
- Small, self-contained GitLab instances for per-person administration training, perhaps on Kubernetes so that a deployment cluster is self-contained as well.
- GitLab Runner implementation patterns, including using platform-specific PaaS.
The primary audiences for and contributors to this information is the GitLab Implementation Eco System which consists of at least:
GitLab Implementation Community:
- GitLab Channel Partners (Integrators)
- Platform Partners
GitLab Internal Implementation Teams:
- Quality / Distribution / Self-Managed
- Professional Services
- Public Sector