Jira Align is getting faster, faster.
Background
Last year, Jira Align set out on a journey to enhance its stability, addressing feedback and bugs more efficiently than ever before, while also streamlining the review process of suggestions through standardization. Since then, we have squashed more bugs than we have in the past, and we are proactively identifying bugs internally at a much quicker rate.
This year, as we highlighted in our public roadmap webinar in January, we are doubling down on observability, reliability, and performance to ensure that we are delivering seamless user experiences across the most important screens.
Now, in this blog post, we want to give a glimpse into our performance improvement work, while also sharing a few key wins that you may already be experiencing as you navigate the product.
In addition to these wins, what we’re most excited about is that we’ll be making Jira Align faster, faster than we ever have now that we implemented the tools, infrastructure, and processes that enable quicker performance gains.
As a result, you can expect more frequent releases of performance improvements than ever before. That is why in the upcoming quarter we will start sharing these wins in our release notes for you all to get more detailed insights into the areas that are improving on a more regular basis.
Performance improvement work
All Jira Align teams are investing in performance improvements. One of our biggest performance improvement workstreams is broken down into three stages, and is currently scoped to three features: program board, backlog, and roadmap. This does not mean that we are only investing in performance improvements in these areas, but it does mean that these areas have an elevated focus. We chose these three features because they are big, important, and oftentimes slow. Once we hit our success metrics, we will move onto other features.
In the first stage of the project, we added instrumentation and conduct analyses. This involves integrating telemetry and logging, conducting comprehensive query decompositions and analyses, as well as assessing the influence of different entities and their respective cardinalities on performance.
We have successfully completed the first stage for each of these features.
In the second stage, we created a plan by identifying the solutions that may solve the biggest problems slowing down these pages. We pitch ideas, talk and learn from other Atlassian teams who have successfully improved their products' performance, and make tradeoff decisions about the most important user experiences.
We have successfully completed the second stage for each of these features.
The magic happens in the third stage, which is when we will ship performance improvements and add monitoring and alerting. We may improve the technology and techniques that make the page faster, add pagination or lazy loading, refactor entire queries, cache, or create newer, quicker libraries that parallelize and compose queries within our repository layer. On top of all that, we are investigating product changes – such as new filtering or administrative options – that help initial load times.
We are in the thick of this stage, and we plan to add public roadmap items for each feature so that you can get more detailed descriptions and timing expectations.
As we continue our work in this final stage, we will consistently deliver performance improvements to these three features to enhance your experience as quickly as possible. As you will see in the next session, some of this performance improvement work is already producing big wins.
Noteworthy performance wins
In general, Jira Align is performing better than it has in the past. As you can see in this graph below, the overall product’s response time has improved over the past couple of weeks for all customers, and we suspect it’ll continue to drop as we ship more improvements.
In addition to this trend, Jira Align has consistently struggled with ‘problem requests,’ or requests that take longer than the average request time. However, over the past couple of months, we have significantly decreased the number of problem requests by breaking down queries, enhancing SQL read locks, and reducing session locking. This will result in users having fewer moments when the app becomes a blocker to progress.
Other performance improvements
| Product area | What shipped? | What's next? |
| API |
Significantly reduced timeout occurrences. Additionally, customers may experience requests responding 75% quicker. |
Cursor-based pagination, which will offer even more latency reductions for customers with many entities.
|
| Program board |
The initial page loads 40% faster following refactoring.
|
Continuous improvements to both the initial load and grid rendering time. |
| Backlog |
The initial page load for the objective, capability, feature, and epic backlogs is up to ~33% faster following session locking improvements. |
Performance improvements for the kanban column view and the list view for features. |
| Forecast | For planning increments with more than 10 programs, the forecast table and the forecast table 'load more' button loads >100% faster following refactoring. | The same improvements at the team level. |
| New navigation |
Following browser caching, new navigation is loading cached data much faster. We are seeing a 400% throughput drop, resulting in features experiencing significantly reduced delays due to new navigation. |
Moving onto other performance areas. |
Was this helpful?
Thanks!
Jake Comito
5 comments