Skip to content

FAQ

This page covers the questions that matter for evaluation, not marketing copy.

Since version: FAQ updated for the 0.9.0 server line, 0.4.0 TypeScript SDK, and v0.1.0 Go SDK.

Caveat: Vaylix is pre-1.0. The answers below describe current behavior, not promises about features that have not shipped.

It is pre-1.0. That means the project is usable, but the storage format and compatibility guarantees are still governed by explicit pre-1.0 policy rather than a 1.x stability promise.

It is already running in production inside AuthSafe. That is evidence of use, not a substitute for your own qualification.

Vaylix uses a Raft-style quorum model. A write acknowledged before the crash has already been committed to a majority of nodes and is not lost. A write appended locally but not quorum-committed can be rolled back during leader replacement.

Can I use Vaylix as a drop-in replacement for Redis?

Section titled “Can I use Vaylix as a drop-in replacement for Redis?”

No. The protocol is different, the client API is different, and the data model is key-value only.

It supports MULTI / EXEC / DISCARD with serialized leader execution and WAL-backed commit.

Current limits:

  • no MVCC
  • no distributed transactions
  • no user-selectable isolation levels

What happens if a node falls behind in replication?

Section titled “What happens if a node falls behind in replication?”

The follower catches up through normal append replication. If it is too far behind, the leader can use snapshot bootstrap and then continue from the snapshot boundary.

Yes. WAL segments, snapshots, and related storage metadata are encrypted at rest through the storage keyring model, which also supports key rotation.

TypeScript and Go. The TypeScript SDK targets Node.js, and the Go SDK is available as github.com/vaylix/vaylix-go.

What is the performance compared to Redis?

Section titled “What is the performance compared to Redis?”

Vaylix fsyncs every acknowledged write and, in HA mode, waits for quorum commit. It is slower than Redis under default configuration. That is the cost of the durability and consensus model.