Confluence vs BookStack

BookStack's shelves, books, chapters and pages hierarchy fits content that has a stable owner and a long shelf life: handbooks, policy libraries, runbooks and onboarding guides. Readers find a policy by walking down a known path, and editors know where a new page belongs. The model becomes harder to live with in project-based organizations that open many loosely related Confluence spaces, each with its own home page, templates and permissions, and that reorganize them every quarter. In BookStack, a page cannot sit outside a book, and moving content between books is a manual operation, so teams that treat a space as a disposable project container will feel the constraint quickly.

Before you decide, export a representative inventory of your Confluence content and map it onto the target model. List every space, its home page and its page tree depth, then note which spaces use labels, which rely on page-tree nesting for meaning, and which contain pages that are referenced from several places. Pages that belong in two spaces need a single chosen home, with links from the other location. Labels and cross-space search do not translate one to one into shelves, so decide in advance whether a label such as a product name becomes a shelf, a chapter or a tag you search on. Confirm with the BookStack documentation how it handles tags and search in your version, because the answer determines how much manual sorting remains.

The restructuring work is usually larger than the data transfer. Plan a pass that merges small or dormant spaces, splits oversized ones into books, and renames chapters so that people searching by team or topic still find content. Assign an owner to each shelf, since BookStack's role and permission system is applied at these levels and an unowned shelf tends to accumulate stale pages. Treat this as a content design exercise with the people who use the wiki daily, and budget the time for it separately from the technical migration, because it is the step most often skipped and most often blamed later.

BookStack publishes official installation scripts for Ubuntu 22.04, 24.04 and 26.04, and it does not publish official Docker images. Its installation documentation points teams to community Docker setups, such as the LinuxServer.io image, instead. That means any container you run is a third-party build, and the first decision is whether your security team accepts that provenance. Before standardizing, check who maintains the image, how often it is rebuilt, whether it is signed or published with build provenance, and whether you can build it yourself from the BookStack release tag. Record the image digest you deploy so an unexpected change is visible.

The stack requirements shape the work regardless of packaging. BookStack needs PHP 8.2 or later and MySQL 8.0 or MariaDB 10.6 or later, so the database version in your environment must be checked against the image's own base. Patch cadence becomes your responsibility: you need a process to track BookStack releases, such as v26.09.1 published on September 29, 2026, and to rebuild or pull updated images when PHP, the database or the application receives a fix. Backups must cover the database, the uploaded files directory and the application configuration, and a restore should be tested on a separate host because a container image alone holds none of your content.

Confirm these points before you commit: the image's maintainer and release process, which PHP and database versions it ships, how it handles upgrades between BookStack releases, and whether it supports the sign-in methods you need, including OIDC, SAML2 and LDAP, and enforced MFA. Ask your platform team who will own the rebuild and the emergency patch path, and whether an official installation on a supported Ubuntu release would be simpler for your organization. Write these answers into your runbook so the choice of image is a documented decision rather than a default.

The Confluence side of the comparison is mostly recurring. The cost is a subscription that scales with the number of users and the edition chosen, and the price for your tier should come from a written quote from Atlassian or a partner, not from a list you assume is current. If you run Data Center, add the server or cluster hardware, the database, load balancers, storage and the staff time needed to patch and upgrade the platform. Confirm the Data Center end dates with your Atlassian account team, since the Atlassian Ascend announcement sets the timeline for new subscriptions and expiry, and any renewal you sign now should be priced against that schedule. Each user tier increase, added instance and extra year of subscription raises the bill, while consolidating sites or trimming inactive accounts lowers it.

BookStack carries no license fee, so its costs are the infrastructure and the people who run it. Plan for a virtual machine or container host, a MySQL 8.0 or MariaDB 10.6 or later database, PHP 8.2 or later, backups of both the database and the uploaded files, and regular patching of PHP, the operating system and the web server. These costs recur monthly or annually, and they rise with the number of users, attachment volume and the availability you require. A paid support plan from the BookStack team is optional. Ask whether a plan is needed for your risk profile and what it covers, because it is the only license-like line item. Also ask whether your team will rely on community Docker setups, since that choice shifts maintenance work onto your staff.

The largest one-time cost on both sides is migration labor. Budget for content mapping into shelves, books and chapters, cleanup of Confluence macros and attachments that do not carry over cleanly, permission design, single sign-on configuration, and user onboarding. Scope this in writing before work begins so the quote covers testing and cutover, not only the import. Set a date for recovering the cost of parallel running, when both systems are live. In a fixed-scope engagement, ask the implementation partner to itemize installation, integration, hardening, backup setup and documentation separately, so you can see which items recur and which end at handover.