InterMine Cloud: Making InterMine cloud-native and easing deployments

GSoC 2019 was fun and I learned a lot from the InterMine Cloud project. In this blog post, I am going to summarise the work that I did on the project. A detailed technical description of all the work done will be published elsewhere.

InterMine is a powerful data warehousing, integration and analysis tool used to store and share genomics data. However, setting up an instance of InterMine is a time consuming and error prone process. It also requires technical knowledge and some familiarity with Java, Postgres, Solr, Perl and shell scripts. These issues create a barrier for entry and friction in adoption of InterMine by the bioinformatics community.
To solve these issues, we went back to the drawing board and spent two months planning and searching for simple and feasible solutions.

So, the first thing that we did was packaging InterMine into Docker containers.

InterMine on Docker

Repo: https://github.com/intermine/docker-intermine-gradle
Commits: https://github.com/intermine/docker-intermine-gradle/commits?author=leoank


Packaging InterMine into Docker containers helped us to reduce required dependencies to set up an InterMine to just two (Docker and Docker Compose). Previously you had to go through tens of pages of InterMine docs to get everything set up and configured correctly to start a new InterMine.

But, packaging InterMine into Docker containers was not a trivial task. Unlike other applications where we can have a single generic container image that can be used by different users, InterMine needs to be custom built for every user. Also, the build requires coordination with other services like Postgres and Solr.

So, instead of having a single Docker image, we now have a set of Docker images that can be orchestrated together to build custom InterMines. These Docker images can be configured easily using environment variables and config files for easier cloud deployments.

Usage instructions for these Docker containers are documented here.

After packaging InterMine in Docker containers, the second thing we did was to write the cloud infrastructure needed for deploying InterMine as Code.

InterMine Cloud Infrastructure as Code

Repo: https://github.com/intermine/intermine-cloud
Commits: https://github.com/intermine/intermine-cloud/commits?author=leoank

To achieve an easy to use and reproducible cloud infrastructure setup and deployments, we used three technologies: Terraform, Kubernetes and Helm.

Terraform is used to define required infrastructure as code. We now have Terraform scripts that can be used to spin up a Kuberenetes cluster on Google Cloud Platform with correct configs in just minutes.

Kubernetes is a production-grade container orchestration platform. It makes easier to manage containers on cloud.

Helm is like a package manager for Kubernetes. We wrote helm charts for deploying single InterMine instances and also entire InterMine Cloud components. Using these charts, users can deploy a custom InterMine in just minutes now.

Doing all this work standardised the cloud deployment process for InterMine. But, we didn’t stopped here though. We took this one step further, which finally brings us to InterMine Cloud.

InterMine Cloud

Repos:
Compose: https://github.com/intermine/intermine_compose
Configurator: https://github.com/intermine/intermine_configurator
Wizard: https://github.com/intermine/wizard

Commits:
Compose: https://github.com/intermine/intermine_compose/commits?author=leoank
Configurator: https://github.com/intermine/intermine_configurator/commits?author=leoank
Wizard: https://github.com/intermine/wizard/commits?author=leoank

InterMine Cloud is a SaaS platform that offers InterMines as a service to its users. It brings a whole new way to use InterMines and makes it accessible to a much larger group of users. We envisioned a completely new user workflow that removes all the technical burden from a user.

InterMine Cloud Workflow

The work we did on InterMine Cloud is completely reusable and we encourage others in to community to host their own InterMine Clouds. The diagram below gives you a brief overview of the architecture.

InterMine Cloud Architecture Overview

InterMine Cloud has four main components:

  • InterMine Compose
  • InterMine Configurator
  • Wizard
  • Kubernetes environment

InterMine Compose

Compose is responsible for authentication, authorisation and building custom InterMines using config files generated by InterMine Configurator. It also acts as a proxy to InterMine Configurator and the underlying kubernetes environment.

InterMine Configurator and Wizard

My mentors wrote configurator and wizard. Together they are responsible for generating a mine config that is used by InterMine Compose. Wizard asks a series of relevant question to the user about the data file, which is then processed by configurator to generate a config.

Kubernetes environment

The underlying Kubernetes environment is a standard Kubernetes cluster with few InterMine cloud specific components added. These specific components includes a Solr service and a distributed shared filesystem enabled by Rook.

Future Work

InterMine cloud is functional but a work in progress. It will take few more weeks to reach alpha. We have planned to add few more features before a public release and also actively looking for community feedback and suggestions.

Advertisements

Published by

ank

I am in the middle of a long journey and I am enjoying it. :) GSoC 2019 student. @leoank on github and gitlab.