Cloud size limitations, considerations, workarounds

[Keith]

Dear GIANT Jira Cloud admins,

How do you pull it off?

Has anyone found limitations in certain sections of cloud function?
We are reaching

• 11 million+ issues
• almost 100k sprints
• 2,000 projects
• 30 million comments
• 150-200 million api requests per month / 5-8 million per day
• 240k saved filters

We continue to have to make adjustments to various sections of Jira, disable features, or create custom solutions to solve long standing known issues internally.

Recently, Atlassian advised that because we are allowed to create as many cloud sites as possible we should split up our 1 production site into as many possible using org, divisions, like minded workflows.

After explaining this is not an option because we tried in the past having to revert the division quickly after hundreds of complaints.

[Nic Brough -Adaptavist-]

We advise our clients to split up their systems, by organisation, silo, division, or whatever the teams are for.  The main thing to do is look at who needs to work with whom.

Your numbers are simply too large for Cloud, it won't work.  You should get an Enterprise licence and split up into several Cloud systems, or get a Data Center licence if you want to carry on running with that many issues, projects, users and API requests in a single system.

We've recently migrated one of Atlassian's largest customers (similar size to yourself) to the Cloud, and we had to get them to "housekeep" 4/5ths of the data, to bring it down to a size where it could be done.  Another one (of half the size) simply gave up on Cloud and converted their server to DC.

[Keith]

Glad to get a quick reply from an expert of your caliber Nic!
Thank you!

Sadly, this is all vital data & the numbers have even been reduced from reality.
With tens of thousands of active users on our current site it is indeed impossible to be a "good steward" our dedicated team with decades of combined Jira experience engage in this battle daily.
Those stats don't even include our yearly recurring efforts to push info into a connected archive site (which is of course also rather large).

Atlassian continues to push us to move to Cloud while at the same time refusing to tell us the bottleneck numbers. The ones we know about have been discovered as we go.

Its hard to align two of the 5 company values of "open company, no bullshit" and "Don't #@!% the customer" with an obvious effort to conceal the real bottlenecks when it comes to Jira at its largest scale.
Without clear answers its hard to even move to cloud if we did split it up because we wouldn't know at what point we would have to split it up again until we started to notice issues.

Maybe we can all continue to push Atlassian to stop trying to put a band-aide on the scale issue and fix the issues with the framework that cause its largest customers to consider alternative solutions instead of moving to cloud.

[Nic Brough -Adaptavist-]

Every number you gave us (except the number of projects, and possibly filters if you could take out the now unused ones) are a bottleneck.

Atlassian's "cloud first" can be too pushy at times, in my opinion.  The numbers you have are screaming for splitting into many Cloud systems, or building a monolithic data-center if you want them all in one place.

Much as I'm an Atlassian fan, and should probably "toe the line", I simply can't say Cloud is for you, unless you're willing to split, or massively reduce your content.  It's far bigger than anything that I know of that has successfully gone to Cloud and it's far bigger than other systems I've seen try and fall back to DC.