Best practices for high availability for Atlassian products

[Ismar_Slomic]

We are designing setup of Jira, Crowd, Bamboo, Crucible, Stash and Confluence in a large enterprise with expected 2000 users every day.

Are there any best practices on how to achieve HA on most critical apps such as Jira, Confluence and Crowd? Also, how is the licensing when for instance setting up clustering?

Thanks!

[Steve Gerstner [bridgingIT]]

You can use a cold standby strategy instead  -> https://www.atlassian.com/licensing/purchase-licensing#licensing-9

I think as the Data Center product line is just a couple of months old, crowd will follow up soon as data center version. But maybe Casper can give some more infos to this point

[Ismar_Slomic]

Thanks, Steve! I´m still wondering how the licensing work for Data Center, 20.000 / per 1.000 user / per application?

[Caspar Krieger]

JIRA, Confluence and Stash have Data Center editions to provide high availability through clustering.

Crowd doesn't have an officially supported high availability solution yet. The feature to vote for is CWD-1053. One of our partners, Go2Group, used to provide a page which illustrated their unofficial, supported-by-them way to cluster Crowd (at http://www.go2group.com/en/services/atlassian-services/atlassian-crowd-a-simple-clustering-solution/) but they seem to have taken that page down; you may like to ask them if they still offer the service.

[Ismar_Slomic]

Thanks Caspar! Couple of follow up questions:

Jira, Confluence and Stash Data Center:

• The price 24.000 / per 1.000 user, does this included all three products or per product?

Crowd clustering:

• As someone already commented in CWD-1053, it is very strange that Atlassian is not supporting clustering for Crowd (which is central component in this architecture). I will try to reach out to go2group!

[Caspar Krieger]

As far as I know, it is indeed $24,000 per 1000 users per product for JIRA, Confluence, and Stash, but you may like to contact our enterprise team ( https://www.atlassian.com/enterprise/contact ) to double check (while we've been using it internally for a while, I think it's a relatively new offering). Re Crowd & clustering, I agree. We are working towards it where we can (I'm a Crowd developer), but we don't have enough certainty to make any promises (yet, at least).