Deploy and Host Microsoft SQL Server on Railway

Microsoft SQL Server is Microsoft's relational database engine: T-SQL, stored procedures, transactions, full-text search and the tooling ecosystem around SSMS, Azure Data Studio and the .NET / JDBC / ODBC drivers.

This template runs SQL Server 2022 on Railway with an engine that actually starts, a generated sa password, a public TCP proxy, and a persistence design built around the one platform limit that stops every other SQL Server listing from working.

About Hosting Microsoft SQL Server

SQL Server is unusually hostile to a container platform, and the reason is the filesystem.

A Railway volume reports a 16384-byte sector size, and SQL Server accepts at most 4096. Any data file placed on the volume is rejected outright:

Cannot use file '/var/opt/mssql/data/x.mdf', because it is on a volume with sector size 16384.
SQL Server supports a maximum sector size of 4096 bytes.

No configuration reaches that check — write-through, trace flag 3979 and trace flag 1800 were each verified applied on a live deploy and the rejection is unchanged, because the check reads the filesystem's reported sector size rather than the IO path.

The consequence is that mounting a volume at /var/opt/mssql, which is the layout every other listing in this category uses, does not give you persistence. It gives you an engine that never starts. On a live control deploy of a competing template, sqlservr crash-loops about once a second with:

/opt/mssql/bin/sqlservr: Error: The system directory [/.system] could not be created.
File: LinuxDirectory.cpp:420 [Status: 0xC0000022 Access Denied errno = 0xD(13) Permission denied]

with zero occurrences of "ready for client connections" in the whole deploy log — while Railway reports the deployment SUCCESS, because no listing in this category publishes a healthcheck. A greenly-deployed database server whose engine died during startup is the normal state of this category.

This template takes the other route. The data files stay on the container's own filesystem, where the sector size is correct and the engine starts normally, and durability is provided by a backup and restore cycle onto the volume:

The engine first starts on a private port Railway does not route, restores the most recent backup of every user database, and replays the recorded SQL logins with their original SIDs and password hashes, so applications keep working without re-granting anything.
Only then does it shut down and re-start on 1433, so the first connection the deploy ever accepts is against restored state.
BACKUP DATABASE ... TO DISK runs on a schedule (MSSQL_BACKUP_INTERVAL_MINUTES, default 15) and once more on shutdown, so an ordinary redeploy loses nothing.

Measured on a live Railway redeploy: a seeded table row and a created login were both present after the container was replaced.

The container also repairs the volume's root ownership, sizes the engine's CPU affinity from the cgroup quota rather than the host's core count, and refuses to boot on an empty sa password.

Why Deploy Microsoft SQL Server on Railway?

Railway is a singular platform to deploy your infrastructure stack. Railway will host your infrastructure so you don't have to deal with configuration, while allowing you to vertically and horizontally scale it.

By deploying Microsoft SQL Server on Railway, you are one step closer to supporting a complete full-stack application with minimal burden. Host your servers, databases, AI agents, and more on Railway.

Compared with running it yourself, you get a managed volume, a generated credential, a public endpoint and a scheduled backup cycle without touching a configuration file.

Common Use Cases

A development or staging database for a .NET application that targets SQL Server in production, reachable from your laptop over the TCP proxy.
Migrating or testing an existing SQL Server schema — restore a .bak, run your migrations, point an ORM at it.
A backend database for an internal tool that already speaks T-SQL, alongside the rest of your services on the same private network.

Dependencies for Microsoft SQL Server

SQL Server 2022 (provided by the image).
A Railway volume, used for backups and login state.

Deployment Dependencies

Upstream image: mcr.microsoft.com/mssql/server, pinned by digest.
Wrapper image and entrypoint: bon5co/mssql-railway
SQL Server documentation: learn.microsoft.com/sql

Implementation Details

Connect from outside Railway with SQLSERVER_URL, and from another service in the same project with SQLSERVER_PRIVATE_URL. SQLSERVER_HOST, SQLSERVER_PORT, SQLSERVER_USER and SQLSERVER_PASSWORD are published separately for drivers that want the parts.

sqlcmd -S , -U sa -P "$MSSQL_SA_PASSWORD" -C -Q "SELECT @@VERSION"

The image ships the Developer edition, which is free but licensed for development and test use only. For production use, set MSSQL_PID=Express — free for production, with a 10 GB limit per database and a lower memory ceiling — or supply your own licence key.

Backups live on the volume under /data/backups and are taken every MSSQL_BACKUP_INTERVAL_MINUTES minutes and on shutdown. A crash between backups can lose up to that interval of writes; lower the value if that matters more than the write cost.