Quarterly Cloud performance and scale update

[#1]

Hi Atlassian Enterprise Community!

Organizations must have products that are performant at scale. Without them, delivering on business objectives can be challenging, as teams are held up completing simple tasks rather than finding more efficient ways to keep work moving. Moreover, teams can’t access all the value available in their products if they can’t use them. That’s why we’ve continued to invest heavily in this space.

With Server and Data Center, you’ve had to make improvements to your infrastructure to mitigate and optimize the performance of your products - and many of you have spent considerable time and resources doing that. For our enterprise customers, we’ve built additional features to help address these needs at scale and will continue to do so for those of you who are unable to move to Cloud at this time.

We know that the expectations you have for your products haven’t changed even though we take on the responsibility of hosting and maintaining the infrastructure our Cloud products run on - ensuring that they’re available, performant, and scalable. We want to assure you that meeting your performance and scale needs is our top priority.

Speeding up Cloud performance

To date, we’ve seen that our investments in performance and scale have already delivered some major wins.

Decreasing page load times

Teams should be able to move seamlessly through their workflows, so we’ve focused on decreasing the amount of time it takes for pages to load - thus keeping teams productive. We evaluate load time based on time to interaction (TTI). Overall, we’ve seen significant improvements in performance over the last year. In the last 5 months alone, we’ve seen initial load times decrease ~14.6% for the Issue view in Jira Software. In Confluence, we’ve seen similar gains - lowering the initial load time of the Page view by ~27%.

Increasing the scalability of our products

We also want everyone in your organization to be able to use their Atlassian products. In Q2 2021, we announced that Jira Software and Confluence Cloud support 20,000 users on a single instance - a 10x gain in just three years! But we’re not stopping there. We’re shipping our 50K EAP in Q3 of 2021 and are targeting support for 35K in the second half of 2022.

We’re proud of the fact that we’ve made performance improvements on top of scaling our Cloud products - not sacrificing one over the other.

Migrating more of our enterprise customers

As a result of these investments, we’re seeing even more enterprise customers migrate to Cloud. Along with this comes new challenges and more complex use-cases that test our boundaries and further drive us to make performance better.

For all customers who are assessing a Cloud migration, we work with you to conduct an L1 assessment so you can see what performance is like based on individuals with similar criteria who have migrated. For larger enterprises, we’ll also run an L2 assessment to dig deeper into your specific data shape.

Not only does this provide more clarity to you, but it also enables us to continue validating and prioritizing the performance improvements we make so that our products continue scaling above and beyond your needs. For example, we recently migrated a large enterprise to Jira Software and Confluence Cloud. They brought with them 10K+ users, 100+ integrations, and API requests rates of 50K at peak load. Our learnings from the migration have enabled us to prioritize increasing the API request rate.

Staying ahead of performance

Because of our hyperfocus on performance, we’ve been able to identify the things that most frequently impact it. Here are some examples:

• Browser configurations and updates

• Web-based security plugins or third-party apps

• Allow-listing Atlassian domains

• VPN

• Using HTTP1 protocols instead of HTTP2

• Running on less than 4 physical cores and having network speed below 10Mbps

In addition, we’ve recently launched a large-scale program spanning Jira Software Cloud and Confluence Cloud to address the performance needs of our enterprise customers and identify additional improvements that we can make. These teams are focused on addressing performance in the most frequently used experiences to ensure they’re on par with or better than they were on Data Center.

We’ll be delivering these improvements in phases. Phase one will be focused on the Issue view in Jira Software and Page view in Confluence because these are the most used. Next, we’ll expand our effort to other key experiences in Jira Software and Confluence to further provide you with a faster, more seamless experience in Cloud.

As we continue to make improvements, we’ll continue to keep you informed with updates on Community and you to follow our progress on the Cloud roadmap. But we also want to hear from you. We encourage you to leave your thoughts in the comments section below and we look forward to continuing to provide you with updates on our progress.

[Darryl Lee]

Thanks for sharing!

[Dane Kantner]

when is it expected that exact text searching of text fields will be fixed?

[Taranjeet Singh]

@Yogesh Singh Thanks for sharing these Cloud performance and scalability updates!

These performance improvement initiatives and programs look promising.

Are there any timelines for when the Phase 1 and Phase 2 of Atlassian Performance Program would be delivered?

Is this Atlassian Performance Program only for Enterprise plans of Cloud or it applies to all Cloud plans?

[Yogesh Singh]

@Taranjeet Singh Phase 1 is expected to land by Q3 of 2022 while Phase 2 is still being planned. 
The performance program has nothing to do with Enterprise or any other plans. Its benefits will be visible to all the customers irrespective of their size or plan. 

[Taranjeet Singh]

Thank you for your responses, @Yogesh Singh !

[Mandy Ross]

Great article, @Yogesh Singh !

[Yogesh Singh]

@Dane Kantner regarding your question "when is it expected that exact text searching of text fields will be fixed?" 
Could you give me more details? What needs fixing here. 

[Dane Kantner]

@Yogesh Singh the functionality to search in JQL with exact text lookups is disabled/broken in the cloud. This issue was submitted 12 years ago as a bug but 2 weeks ago the Product team changed it to a feature request.

You can only do fuzzy ~ searches that do not retrieve by exact text b/c of word stemming and the underlying engine. It is purported that enabling exact text searching of fields in the cloud would be too resource intensive on the cloud compute that is shared among tenants. To work around this, we end up having to duplicate all of the fields we want to exact text search on and hash them in new custom fields with SHA and search by the SHA hash instead (in automations that use the API), or we duplicate the fields and remove spacing/punctuation/etc. in the duplicate fields.

Here is the issue for it:

https://jira.atlassian.com/browse/JRACLOUD-21372

[SG]

@Dane Kantner, thanks for reaching out. I can confirm that we are already working on fixing this issue.

It is somewhat complex, so unfortunately I can't share an ETA at the moment. We will share updates on the jira.atlassian.com ticket whenever there is anything significant to report.

- Syed

[Srinatha T]

Thanks for sharing. 

[Karthick S]

Thanks for sharing these statistics @Yogesh Singh 

[Mike]

@Yogesh Singh @Karthick S @Srinatha T @SG @Dane Kantner  The article says single instance can handle 20k users, could you please share a little more on this topic? is this a cloud server your are referrring too? if yes, then what's the configuration? 

- Raj

[Srinatha T]

@Mike you can raise your query here. Our sales team will be more than happy to answer your questions. 

[Yatish Madhav]

Thanks for this article @Yogesh Singh - we are constantly looking into ways in which we as the end consumer can improve the performance of Jira Cloud as well. We have taken over an instance that has around 800 projects and numerous "behind the scenes" elements like schemes, statuses, issue types, etc and we would like to clean that up so that using Jira becomes a click of the finger type of work! :)