Since March 30, 2026, first-time buyers can't get Jira on Data Center at all. Current customers lose new subscriptions, expansions, upgrades and new Marketplace apps on March 30, 2028, but same-size renewals continue until end of life on March 28, 2029, when Jira Software and Jira Service Management Data Center instances become read-only. Most teams should plan a move to Atlassian Cloud now; teams whose Jira is heavily customized, must keep data in their own database, or run a core business workflow on it should also price owning that piece.
Atlassian sets out the schedule in its Data Center End of Life FAQ and first announced it on September 8, 2025. Every deadline is 23:59 Pacific on the listed day. For a Jira admin, the second date matters most, because it is when your user tier stops being able to grow.
Which products are affected. Outside the program: Jira Align, plus Bitbucket DC and Bamboo DC, which run past 2029 on a Bitbucket Hybrid License that also includes Bitbucket Cloud. Inside it, for a Jira shop: Jira Software DC and Jira Service Management DC, Crowd DC if your logins depend on it, the mobile apps for DC, and Marketplace apps installed on any of those (Confluence DC is in scope as well). Atlassian will grant extended maintenance or multi-year extensions past end of life only "by exception", undated, so plan as if neither exists. Public-sector teams waiting on FedRAMP High or IL5 get an Atlassian commitment with no date attached; Government Cloud today is authorized at FedRAMP Moderate. For your wiki, see the Confluence Data Center end-of-life page.
Atlassian offers two of them (Cloud in Commercial, Government or Isolated form, or Data Center until 2029); the third, owning some or all of what Jira does, is ours to describe.
One Cloud input to ask your Atlassian account team about: Atlassian says Cloud Enterprise buyers who sign by June 2027 may qualify for a first-year discount. It is Atlassian's program with Atlassian's terms; count it in the Option A column and nowhere else.
Atlassian's official pages show no Data Center price we could confirm, so every input below comes out of your procurement file. Stay: the Jira DC renewal on your last order form, the ScriptRunner, Tempo, Xray or other app renewals listed beside it, and the cost of the nodes, shared home storage, database and Jira admin hours, counted only up to March 28, 2029. Cloud: the per-user subscription for your tier (Atlassian's Jira pricing page or a quote) plus Cloud app fees, over 36 months. Own: engineers × build months × a monthly rate starting at $7,000, the old platform's fees during the build, and a yearly run cost that is your assumption to edit.
Illustration only. Each row starts from a made-up yearly platform figure (not an Atlassian price) standing in for your current DC total or your Cloud quote. Headcount, build months and run costs are our assumptions, and the owned total carries the old platform's fees until the build finishes.
Illustration, not a quote and not an Atlassian price. Example platform costs, team sizes and run costs are assumptions; replace them with your renewal, your Cloud quote and your own estimate.
Reading the rows: a crossing is the first month in which cumulative spend on the owned system is no higher than cumulative spend on the platform. The $48,000 row only ties in month 36, so a single slipped month sinks it; at that spend, Cloud or a same-size DC renewal is the sensible call. The two bigger rows cross in months 17 and 19, but only because the assumed build replaces all of Jira. Under the middle path you rebuild one project, keep Cloud seats for the rest, and both columns shrink. Remember that DC spend stops being an option at the end of March 2029, well short of 36 months, and that test-management or time-tracking apps are often paid from other budgets; add them back. The renewal vs custom build calculator takes your real figures. These rows locate a crossing under stated inputs; they are not a savings forecast.
Export the list of projects with issue counts and last activity, then each project's workflow scheme, issue types, screens and custom fields. Count Marketplace apps and record what each does: scripted fields, listeners, approval add-ons, test management, time tracking. Export automation rules too: in Jira Data Center, a Jira administrator can download all global and project rules as one JSON file. Sort projects into three groups: archive, move to Cloud, rebuild as owned software. Exit check: every active project is in a group, and every app has a named owner who agrees it can go or knows what replaces it.
On Data Center you control the database, which helps. Atlassian's backup documentation says a complete Jira backup involves both the database and the shared home directory, and the automatic service writes an XML snapshot of database contents as a zip in the shared home's export folder. For a clean extraction, pull issues through the Jira Data Center REST API rather than reading tables directly, since the schema is internal. Authenticate with personal access tokens (Jira Software 8.14 and later, Jira Service Management 4.15 and later), sent as a bearer token. If an admin has turned on rate limiting, it applies to external REST calls, so agree an allowance for the export user. Flat CSV exports are capped by search-view limits that default to 1,000 issues; Atlassian's Data Center guide to CSV export in batches pages with tempMax and start and warns that raising the limit can run Jira out of memory. Copy the shared home directory as well, and record checksums for every attachment you move. Exit check: issue, comment, attachment and worklog counts in staging match the source per project.
Service desks need more than issues. The Jira Service Management Data Center REST API exposes customer requests, request types, request attachments, customers, organizations and approvals, which is how you keep who asked, on whose behalf, and who approved. For Assets, Atlassian's object schema export can include objects, attachments and avatars, but not automation and import configurations, object history and comments, or connected Jira tickets. Rebuild the issue-to-object links from the custom field values on the issues themselves. Exit check: a sample of requests shows the right customer, organization, approval decisions and linked assets in the new store.
Model only what the rebuilt workflow needs: work items, a status table driven by an explicit state machine, typed custom fields as real columns where they are queried and JSONB where they aren't, comments, attachments in object storage with a metadata table, worklogs, links between items, and an append-only history table for every field change. Keep the original Jira issue key on each migrated row, and import Jira's change history so old audit questions still have answers. Exit check: the ten reports people actually run today come straight out of SQL views.
Start with the project that justified owning in the first place. Re-implement its transitions, validators and post-functions as tested code; scripted fields and listeners from apps become ordinary functions with unit tests. Imported automation rules are a specification here, not something to port line by line. Ship the new screens to a single pilot team, switched on per user. Exit check: that team completes a full cycle of real work, from creation to close, without opening Jira.
Crowd Data Center reaches end of life on the same date, so any Jira that signs users in through Crowd needs a new login path as part of this work. Put SSO through your identity provider in front of the new system, map Jira groups to roles, and decide where Cloud users and owned-system users are provisioned from. Exit check: joiners and leavers flow from the identity provider to every system without anyone touching Crowd.
Run old and new side by side for at least one full work cycle, with a nightly job comparing open counts, status distribution and SLA results per project. Move teams one at a time, set the matching Jira projects to read-only through permission schemes, and keep the final export in your own storage. Leave months of slack ahead of March 28, 2029, and after that date do not leave the read-only instance exposed to the internet. Exit check: no active work is created in Data Center for a month, and old issue keys resolve to archived records.
On a Jira exit, our AI pod teams hand coding assistants the tedious work: field-by-field mapping of customfield IDs to columns, test cases generated from exported automation JSON, and first drafts of paging and reconciliation scripts. A person reviews each change, and nothing merges until tests pass.
Roles for a Jira exit: a tech lead who decides which projects go to Cloud and owns schema and cutover; one, two or three senior full-stack engineers on the rebuilt workflow and its UI; a data engineer for the REST and CSV export when the instance holds years of issues and a heavy shared home (hire data engineers), and a platform engineer for SSO, Crowd retirement and hosting (hire DevOps engineers). QA, part-time, signs off the nightly diffs. Many teams begin with a pod of 3 to 5 senior engineers (a tech lead plus engineers), and a full pod is about seven. Rebuilding a single service desk or project under the middle path can begin with a tech lead and a couple of engineers, and the same team can run the Cloud migration for the rest with our cloud migration engineers.
Ryz engineers on your team start from $7,000 per month (our published range is $7,000–$15,000 per engineer per month, mostly mid-level to senior), quoted per team. If the cost frame says one Jira workflow is worth owning, send us your project list and scope your team from it.
Yes, provided the renewal keeps your current user tier and app list. What ends on that date is growth: new subscriptions, expansions, upgrades and new Marketplace DC apps. Renewing like-for-like, or stepping down, stays open until March 28, 2029, and Atlassian prorates any annual term that would extend beyond that day.
The license and its apps lapse, and Jira freezes: existing issues stay readable, but no new data can be added and no new actions taken. Atlassian stops updates and support and suggests taking such an instance off the internet. Anything beyond that day is an Atlassian exception with no published date.
No. Your Git hosting and build server keep running on Data Center after 2029, and Atlassian gives existing customers a Bitbucket Hybrid License spanning Bitbucket DC, Bamboo DC and Bitbucket Cloud. Jira Align also sits outside the program. The products that do reach end of life are the Jira and Confluence Data Center lines, Crowd, the DC mobile apps and Marketplace apps for them.
The drivers are how many projects you rebuild rather than move to Cloud, how much logic lives in scripted apps and automation rules, the size of the attachment store, and whether Crowd must be retired at the same time. A week-one inventory and an export dry run are what make a realistic plan possible, and a team-by-team cutover keeps each step small.
For plain Scrum and Kanban projects, almost never; Cloud usually wins there. For a heavily scripted operations project it can be. Set 36 months of your Cloud quote against headcount times build months times a rate from $7,000, the DC or Cloud fees paid while building, and your own run-cost estimate, then see whether the lines meet.
Questions we didn't answer? Email info@ryzlabs.com.
Thanks — your message has been sent. We’ll get back to you soon.
Something went wrong while sending your message. Please try again or email info@ryzlabs.com.