rqlite is a lightweight, distributed relational database, which uses SQLite as its storage engine. Forming a cluster is very straightforward, it gracefully handles leader elections, and tolerates failures of machines, including the leader. rqlite is available for Linux, OSX, and Microsoft Windows.
Check out the rqlite FAQ.
rqlite gives you the functionality of a rock solid, fault-tolerant, replicated relational database, but with very easy installation, deployment, and operation. With it you've got a lightweight and reliable distributed relational data store. Think etcd or Consul, but with relational data modelling also available.
You could use rqlite as part of a larger system, as a central store for some critical relational data, without having to run larger, more complex distributed databases.
Finally, if you're interested in understanding how distributed systems actually work, rqlite is a good example to study. Much thought has gone into its design and implementation, with clear separation between the various components, including storage, distributed consensus, and API.
rqlite uses Raft to achieve consensus across all the instances of the SQLite databases, ensuring that every change made to the system is made to a quorum of SQLite databases, or none at all. You can learn more about the design here.
- Trivially easy to deploy, with no need to separately install SQLite.
- Fully replicated production-grade SQL database.
- Production-grade distributed consensus system.
- An easy-to-use HTTP(S) API, including leader-redirection and bulk-update support. A command-line interface is also available, as are various client libraries.
- Discovery Service support, allowing clusters to be dynamically created.
- Extensive security and encryption support, including node-to-node encryption.
- Choice of read consistency levels.
- Optional read-only (non-voting) nodes, which can add read scalability to the system.
- A form of transaction support.
- Hot backups.
The quickest way to get running on OSX and Linux is to download a pre-built release binary. You can find these binaries on the Github releases page. If you prefer Windows you can download the latest build here. Once installed, you can start a single rqlite node like so:
rqlited -node-id 1 ~/node.1
-node-id isn't strictly necessary at this time, but highly recommended. It makes cluster management much clearer.
This single node automatically becomes the leader. You can pass
rqlited to list all configuration options.
Alternatively you can pull the latest release via
docker pull rqlite/rqlite.
Forming a cluster
While not strictly necessary to run rqlite, running multiple nodes means you'll have a fault-tolerant cluster. Start two more nodes, allowing the cluster to tolerate failure of a single node, like so:
rqlited -node-id 2 -http-addr localhost:4003 -raft-addr localhost:4004 -join http://localhost:4001 ~/node.2 rqlited -node-id 3 -http-addr localhost:4005 -raft-addr localhost:4006 -join http://localhost:4001 ~/node.3
This demonstration shows all 3 nodes running on the same host. In reality you probably wouldn't do this, and then you wouldn't need to select different -http-addr and -raft-addr ports for each rqlite node.
With just these few steps you've now got a fault-tolerant, distributed relational database. For full details on creating and managing real clusters, including running read-only nodes, check out this documentation.
There is also a rqlite Discovery Service, allowing nodes to automatically connect and form a cluster. This can be much more convenient, allowing clusters to be dynamically created. Check out the documentation for more details.
Let's insert some records via the rqlite CLI, using standard SQLite commands. Once inserted, these records will be replicated across the cluster, in a durable and fault-tolerant manner. Your 3-node cluster can suffer the failure of a single node without any loss of functionality or data.
$ rqlite 127.0.0.1:4001> CREATE TABLE foo (id INTEGER NOT NULL PRIMARY KEY, name TEXT) 0 row affected (0.000668 sec) 127.0.0.1:4001> .schema +-----------------------------------------------------------------------------+ | sql | +-----------------------------------------------------------------------------+ | CREATE TABLE foo (id INTEGER NOT NULL PRIMARY KEY, name TEXT) | +-----------------------------------------------------------------------------+ 127.0.0.1:4001> INSERT INTO foo(name) VALUES("fiona") 1 row affected (0.000080 sec) 127.0.0.1:4001> SELECT * FROM foo +----+-------+ | id | name | +----+-------+ | 1 | fiona | +----+-------+
rqlite replicates SQLite for fault-tolerance. It does not replicate it for performance. In fact performance is reduced somewhat due to the network round-trips.
Depending on your machine (particularly its IO performance) and network, individual INSERT performance could be anything from 10 operations per second to more than 200 operations per second. However, by using the bulk API, transactions, or both, throughput will increase significantly, often by 2 orders of magnitude. This speed-up is due to the way Raft and SQLite work. So for high throughput, execute as many operations as possible within a single transaction.
By default rqlite uses an in-memory SQLite database to maximise performance. In this mode no actual SQLite file is created and the entire database is stored in memory. If you wish rqlite to use an actual file-based SQLite database, pass
-on-disk to rqlite on start-up.
Does using an in-memory database put my data at risk?
Since the Raft log is the authoritative store for all data, and it is written to disk by each node, an in-memory database can be fully recreated on start-up. Using an in-memory database does not put your data at risk.
- Only SQL statements that are deterministic are safe to use with rqlite, because statements are committed to the Raft log before they are sent to each node. In other words, rqlite performs statement-based replication. For example, the following statement could result in a different SQLite database under each node:
INSERT INTO foo (n) VALUES(random());
- Technically this is not supported, but you can directly read the SQLite under any node at anytime, assuming you run in "on-disk" mode. However there is no guarantee that the SQLite file reflects all the changes that have taken place on the cluster unless you are sure the host node itself has received and applied all changes.
- In case it isn't obvious, rqlite does not replicate any changes made directly to any underlying SQLite file, when run in "on disk" mode. If you change the SQLite file directly, you will cause rqlite to fail. Only modify the database via the HTTP API.
- SQLite dot-commands such as
.tablesare not directly supported by the API, but the rqlite CLI supports some very similar functionality. This is because those commands are features of the
sqlite3command, not SQLite itself.
Status and Diagnostics
You can learn how to check status and diagnostics here.
Backup and restore
You can learn about securing access, and restricting users' access, to rqlite here.
There is a Google Group dedicated to discussion of rqlite.
How do I pronounce rqlite? For what it's worth I try to pronounce it "ree-qwell-lite". But it seems most people, including me, often pronouce it "R Q lite".