Redis storage

In-memory, replicated, blisteringly fast storage. Pairs with the Redis backplane for an all-Redis OpStream deployment.

Install

dotnet add package OpStream.Server.Storage.Redis

Setup

builder.Services
    .AddOpStream()
    .UseRedisStorage(options =>
    {
        options.ConnectionString = builder.Configuration.GetConnectionString("Redis")!;
        options.KeyPrefix        = "opstream:";   // default
    });

// Or
builder.Services.AddOpStream().UseRedis(connStr);

Key layout

Pattern	Type	Purpose
opstream:ops:{docId}	Stream	Append-only op log; entry ids are revision numbers.
opstream:snap:{docId}	String (binary)	Latest snapshot blob.
opstream:snap:{docId}:meta	Hash	revision, timestamp.

Streams give us O(1) appends and O(N) range reads from a base revision, which matches IDocumentStore's access pattern exactly.

Durability

Redis is in-memory by default. For production, enable AOF (appendonly yes, appendfsync everysec) on the OpStream Redis instance, or use a managed offering with built-in persistence (Elasticache, MemoryDB, Upstash, Azure Cache for Redis Premium).

The snapshot mechanism also acts as a compaction strategy: with regular snapshots, you can XTRIM old op-log entries past the snapshot point without losing the ability to rehydrate.

When to pick this

• Latency-sensitive workloads (real-time collaboration is the textbook case).
• You're already running Redis (and now you can use it as both backplane and storage).
• You require ACID across multiple documents — use a relational backend.
• You need cold archive — pair Redis with a tiered archive job to your RDBMS / object store.