The Devhub made the switch to Algolia from a client-side search engine early January 2020. There were several motivations to make the switch:
- Client Side Search had limitations in functionality
- It was difficult to gather metrics from search without having to write custom wrappers around the api
- There was a limit to how large a client side index could get before it'd bog down a user's browser
After doing some research, Algolia was was chosen because:
- the documentation was amazing
- it was very easy to integrate into our application via a
- the possibility of obtaining an open source licence (more on this later)
- the dashboard for Algolia is very nice and easy to use
Algolia is a paid service that has different service offerings based on your team's needs.
The Devhub has an Algolia Open Source Licence. This is a free licence that allows us to utilize a small set of Algolia's service offerings - primarily search. The licence is limited and only allows a
certain number of
search queries per month. These are some factors that your team must be cognoscente of when looking to apply for an Open Source Licence.
Overall the onboarding process to Algolia was seamless and easy. Besides their great documentation, the Algolia team was very helpful and open to communicate with the Devhub Team.
Algolia is one of the moving pieces in Devhub's machinery. Just like all of Devhub's machinery, it is maintained automatically via a pipeline.
Specifically, the management of the search index is automated.
The search index follows the same life cycle as the application code. As the application code is shifted between environments (from dev to test to prod) so is the search index. The Devhub's application life cycle is
[Pull Request](https://help.github.com/en/github/collaborating-with-issues-and-pull-requests/about-pull-requests) based. As new features or bug fixes are released, they are given their own individual development environment for testing prior to integrating to the test and production environments. Similarly, when a new feature is rolled out in the Devhub, a brand new index in Algolia is built at the same time as well.
In addition to Indices being created automatically, the evolution of the index and any of the improvements we make to it (by changing settings/configuring synonyms for example) are codified and pushed to Algolia automatically.
The Algolia Analytics Feature has been immensely helpful in identifying search issues and helping us resolve them.
When Algolia was first integrated, common themes were quickly identified for missed searches. For example, several users were searching for the terms persistent or pvc. These terms would return zero search results. Many of the documents that are available in the Devhub were more specific then those generic search terms and covered topics such as netapp, gluster, or block storage. Synonyms were added to connect the dots between these terms. A few weeks later a drop of more then 25% in no search results was observed! In addition, cross referencing our analytics tool found that the user bounce rate for pages had decreased as well.
Algolia also provides a list of the top searched terms. This has been useful to identify gaps in content that may be missing in the Devhub.
Integrating Algolia has overwhelmingly been a success. In addition, having some analytics available allows insights into the user experience of the Devhub. With this information, the team is able to make more informed decisions on how to best improve the product. The most challenging issue when integrating Algolia was getting started. It took several weeks to acquire the Open Source License. This was the largest blocker but after acquiring it, the integration was fairly seamless.