marmot

Marmot

A distributed SQLite replicator.

What & Why?

Marmot can give you a solid replication between your nodes as Marmot builds on top of fault-tolerant NATS, thus allowing robust recovery and replication. This means if you are running a read heavy website based on SQLite you should be easily able to scale it out by adding more SQLite replicated nodes. SQLite is a probably the most ubiquitous DB that exists almost everywhere, this project aims to make it even more ubiquitous for server side applications by building a masterless replication layer on top.

What is the difference from others?

There are a few solutions like rqlite, dqlite, and LiteFS etc. All of them either are layers on top of SQLite (e.g. rqlite, dqlite) that requires them to sit in the middle with network layer in order to provide replication; or intercept phsycial page level writes to stream them off to replicas. In both cases they are mostly single primary where all the writes have to go, backed by multiple replicas that can only be readonly.

Marmot on the other hand is born different. Instead of being single primary it is "masterless", instead of being strongly consistent, it's eventually consistent, does not require any changes to your application logic for reading/writing. This means:

• You can read and write to your SQLite database like you normally do.
• You can write on any node! You don't have to go to single master for writing your data.
• As long as you start with same copy of database, all the mutations will eventually converge (hence eventually consistent).

Marmot is a CDC (Change Data Capture) pipeline running top of NATS. It can automatically confgure appropriate JetStreams making sure those streams evenly distribute load over those shards, so scaling simply boils down to adding more nodes, and rebalancing those JetStreams (To be automated in future versions).

Dependencies

Starting 0.4+ Marmot depends on nats-server with JetStream support. Instead of building an in process consensus algorithm, this unlocks more use-cases like letting external applications subscribe to these changes and build more complex use-cases around their application needs.

Production status

• v0.4.x is production ready.
• v0.3.x is deprecated, and unstable. DO NOT USE IT IN PRODUCTION.

Features

• Built on top of NATS, abstracting stream distribution and replication
• Bidirectional replication with almost masterless architecture
• Ability to snapshot and fully recover from those snapshots
• SQLite based log storage

To be implemented for next GA:

• Command batching + compression for speeding up bulk load / commit commands to propagate quickly
• Database snapshotting and restore for cold-start and out-of-date nodes

Running

Build

go build -o build/marmot ./marmot.go

Make sure you have 2 SQLite DBs with exact same schemas (ideally exact same state):

nats-server --jetstream
build/marmot -nats-url nats://127.0.0.1:4222 -node-id 1 -db-path /tmp/cache-1.db
build/marmot -nats-url nats://127.0.0.1:4222 -node-id 2 -db-path /tmp/cache-2.db

Demos

Demos for v0.4.x:

• Scaling Pocketbase with Marmot
• Scaling Keystone 6 with Marmot

Demos for v0.3.x (Legacy) with PocketBase v0.7.5:

• Scaling Pocketbase with Marmot
• Scaling Pocketbase with Marmot - Follow up

CLI Documentation

Marmot picks simplicity, and lesser knobs to configure by choice. Here are command line options you can use to configure marmot:

• cleanup - Just cleanup and exit marmot. Useful for scenarios where you are performing a cleanup of hooks and change logs. (default: false)
• db-path - Path to DB from which all tables will be replicated (default: /tmp/marmot.db)
• node-id - An ID number (positive integer) to represent an ID for this node, this is required to be a unique number per node, and used for consensus protocol. (default: random number)
• nats-url - URL string for NATS servers, it can also point to multiple servers as long as its comma separated (e.g. nats://user:pass@127.0.0.1:4222 or nats://user:pass@host-a:4222, nats://user:pass@host-b:4222)
• max-log-entries - Number of change log entries that should be caped in NATS JetStream. This property only applies when stream has not been created. If stream was never created before Marmot with automatically created stream with name prefix stream-prefix with attached subjects prefixed by subject-prefix (default: 1024). It's recommended to use a good high value in production environments.
• log-replicas - Number of replicas for each change log entry committed to NATS (default: 1). Set this value to at least quorum of cluster nodes to make sure that your replication logs are fault-tolerant.
• subject-prefix - Prefix for subject over which change logs will be published. Marmot distributes load by sharding change logs over given shards. Each subject will have pattern <subject-prefix>-<shard-id> (default: marmot-change-log).
• stream-prefix - Prefix for JetStream names for against which each sharded subject is attached. This allows to distribute these JetStream leaders among nodes in cluster. Each stream will have pattern of <stream-prefix>-<shards>-<shard-id> (default: marmot-changes).
• shards - Number of shards over which the database tables replication will be distributed on. It serves as mechanism for consistently hashing JetStream from Hash(<table_name> + <primary/composite_key>). This will allow NATS servers to distribute load and scale for wider clusters. Look at internal docs on how these JetStreams and subjects are named (default: 8).
• verbose - Specify if system should dump debug logs on console as well. Only use this for debugging.

For more details and internal workings of marmot go to these docs.

Limitations

Right now there are a few limitations on current solution:

• You can't watch tables selectively on a DB. This is due to various limitations around snapshot and restore mechanism.
• WAL mode required - since your DB is going to be processed by multiple process the only way to have multi-process changes reliably is via WAL.
• Downloading snapshots of database is still WIP. However, it doesn't affect replication functionality as everything is upsert or delete. Right snapshots are not restore, or initialized.
• Marmot is eventually consistent. This simply means rows can get synced out of order, and SERIALIZABLE assumptions on transactions might not hold true anymore.

FAQs & Community

• For FAQs visit this page
• For community visit our discord or discussions on GitHub

Our sponsor

Last but not least we will like to thank our sponsors who have been supporting development of this project.