R Tyler Croy
74feb6172a
|
||
|---|---|---|
| agents | ||
| apispec | ||
| assets | ||
| auctioneer | ||
| config | ||
| contrib | ||
| eventbus | ||
| examples | ||
| grammar | ||
| rfc | ||
| scripts | ||
| .gitignore | ||
| CODE_OF_CONDUCT.md | ||
| Cargo.lock | ||
| Cargo.toml | ||
| LICENSE.txt | ||
| Makefile | ||
| Procfile | ||
| README.adoc | ||
| dredd.yml | ||
| eventbus.yml | ||
| system.dot | ||
| system.png | ||
README.adoc
Otto
Meet Otto, your friendly continuous delivery companion. Otto does not aim to be the center of the entire continuous delivery process, but rather seeks to interoperate seamlessly with all the various components which make CD work for you.
Problems to Solve
Below is an incomplete listing of the problems Otto aims to solve for:
-
Existing tools to not model the entire continuous delivery process. Using an external tool such as Puppet, or relying on an external provider such as AWS ECS, there can be a "black hole" in the deployment. A point where control is delegated to an external system, and the orchestration tool (Otto), loses sight of what is happening.
-
Expecting "one single instance" to be the hub is unrealistic. Many deployment processes have "development" operated components, and "ops" operated components, with little to no automated hand-off of control between the two.
-
Mixing of management and execution contexts causes a myriad of issues. Many tools allow the management/orchestrator process to run user-defined workloads. This allows breaches of isolation between user-defined workloads and administrator configuration and data.
-
Non-deterministic runtime behavior adds instability. Without being able to "explain" a set of operations which should occur before runtime, it is impossible to determine whether or not a given delivery pipeline is correctly constructed.
-
Blending runtime data and logic with process definition confuses users. Related to the problem above, Providing runtime data about the process in a manner which is only accessible in the delivery process itself, overly complicates the parsing and execution of a defined continuous delivery process.
-
Modeling of the delivery process is blurred with build tooling. Without a clear separation of concerns between the responsibility of build tools like GNU/Make, Maven, Gradle, etc and the continuous delivery process definition, logic invariably bleeds between the two.
-
Opinionated platform requirements prevent easy usage across different environments. Forcing a reliance on containers, or a runtime like the Java Virtual Machine results in burdensome system configuration before starting to do the real work of defining a continuous delivery process. Without gracefully degrading in functionality depending on the system requirements which are present, users are forced to hack around the platform requirements, or spent significant worrying about and maintaining pre-requisites.
-
Many tools are difficult to configure by default. For most application stacks, there are common conventions which can be easily prescribed for the 80% use-case. Ruby on Rails applications will almost all look identical, and should require zero additional configuration.
-
Secrets and credentials can be inadvertently leaked. Many tools provide some ability to configure secrets for the continuous delivery process, but expose them to the process itself in insecure ways, which allow for leakage.
-
Extensibility must not come at the expense of system integrity. Systems which allow for administrator, or user-injected code at runtime cannot avoid system reliability and security problems. Extensibility is an important characteristic to support, but secondary to system integrity.
-
Usage cannot grow across an organization without user-defined extension. The operators of the system will not be able to provide for every eventual requirement from users. Some mechanism for extending or consolidating aspects of a continuous delivery process must exist.
Modeling Continuous Delivery
Some characteristics of a continuous delivery process model which Otto must ensure:
-
Deterministic ahead-of-time. Without executing the full process, it must be possible to "explain" what will happen.
-
External interactions must be model-able. Deferring control to an external system must be accounted for in a user-defined model. For example, submitting a deployment request, and then waiting for some external condition to be made to indicate that the deployment has completed and the service is now online. This should support both an evented model, wherein the external service "calls back" and a polling model, where the process waits until some external condition can be verified.
-
Branching logic, a user must be able to easily define branching logic. For example, a web application’s delivery may be different depending on whether this is a production or a staging deployment.
-
Describe, though not fully, environments. All applications have at least some concept of environments, whether it is a web application’s concept of staging/production, or a compiled asset’s concept of debug/release builds.
-
Safe credentials access, credentials should not be exposed to in a way which might allow the user-defined code to inadvertently leak the credential.
-
Caching data between runs must be describable in some form or fashion. Taking Maven projects as an example, where successive runs of
mvnon a cold-cache will result in significant downloads of data, whereas caching~/.m2will result in more acceptable performance. -
Refactor/extensibility support in-repo or externally. Depending on whether the source repository is a monorepo, or something more modular. Common aspects of the process must be able to be templatized/parameterized in some form, and shared within the repository or via an external repository.
-
Scale down to near zero-configuration. the simplest model possible should simply define what platform’s conventions to use. With Rails applications, many applications are functionally in the same with their use of Bundler, Rake, and RSpec.