Skip to content

For Redis

Redis or Valkey on dedicated RAM, with root access

Redis or Valkey on dedicated RAM, with AOF on local NVMe, RDB snapshots in S3-compatible storage and replicas over a free private network. The cache and queue you already use, on your own VM.
Deploy a VMRead the compute scripts docs

Free to sign up. No egress fees on any product.

Why Redis on Edge

One small VM does a cache node's job

Redis gets its own memory and vCPUs, and you keep root on the box.

  • Real RAM, real CPU

    Redis gets dedicated vCPUs and memory, not a share of a noisy host. Predictable latency without the managed-cache premium.

  • AOF locally, RDB to S3

    AOF on local NVMe for durability, plus scheduled RDB snapshots in Edge Storage for disaster recovery.

  • Replicas and Sentinel

    Replicas on separate VMs over a private network, with Sentinel handling failover. Standard Redis HA, nothing proprietary.

  • Cluster mode when you need it

    Shard a large keyspace across several VMs with Redis Cluster. Most workloads never outgrow one primary.

  • Redis or Valkey

    The library script installs Redis 7. Valkey is a drop-in swap if the licence matters to you; same clients, same commands.

  • Private by default

    Password auth from the script, bound to the private interface, and 6379 open only to your network range.

Reference architecture

Primary and replica, watched by three sentinels

A primary with plenty of RAM, a replica and Sentinel for failover, and snapshots in a bucket. Apps reach it over the private network.
  • Compute

    Runs Redis (or Valkey) on a VM with dedicated RAM, plus a Sentinel process.

  • Compute

    Optional replicas on further VMs, kept in sync over the private network.

  • Storage

    Holds RDB snapshots for disaster recovery and point-in-time restores.

  • DNS

    A cache. record for the primary's private address, if you'd rather not hard-code IPs.

Deploy

A replicated Redis from the library script

The Redis library script handles install, password auth and AOF. The rest is a few lines of redis.conf.
  1. 01

    Create the primary with the Redis script

    Leave redis_password out and one is generated. Attach the VM to your app's private network afterwards.

    shell
    $ edge compute create --name cache-1 --size s-2vcpu-4gb \
    --image ubuntu-24 --region london --script redis \
    --script-param redis_password="$(openssl rand -hex 24)"
  2. 02

    Bind it to the private network

    Use noeviction instead of allkeys-lru if Redis also holds queues: an error is better than a silently dropped job.

    /etc/redis/redis.conf
    bind 127.0.0.1 10.185.0.4
protected-mode yes
requirepass <password>
appendonly yes
appendfsync everysec
maxmemory 3gb
maxmemory-policy allkeys-lru
  3. 03

    Add a replica and Sentinel

    Create cache-2 the same way. Run a sentinel on cache-1, cache-2 and one app VM so any two can agree on a failover.

    redis.conf, sentinel.conf
    # cache-2: /etc/redis/redis.conf
replicaof 10.185.0.4 6379
masterauth <password>

# sentinel.conf on three VMs
sentinel monitor cache 10.185.0.4 6379 2
sentinel auth-pass cache <password>
sentinel down-after-milliseconds cache 5000
  4. 04

    Snapshot RDB to Edge Storage

    Run it from cron and point it at the replica: redis-cli --rdb pulls a fresh snapshot over the network, so the primary never forks for a backup.

    shell
    $ redis-cli -h 10.185.0.6 -a "$PW" --rdb /tmp/dump.rdb
$ aws s3 cp /tmp/dump.rdb \
    s3://redis-snapshots/$(date +%F).rdb \
    --endpoint-url https://storage.edge.network

Prefer to hand it off? Give the job to your AI agent or have our engineers do it.

What it costs

Primary, replica and Sentinel for under $40

Two Standard VMs and a bucket. No per-node-hour cache premium and no charge for replication traffic.
  • No per-hour cache node premium
  • Replication traffic on the private network is free
  • Redis or Valkey at the same price
  • Billed hourly from a prepaid balance
See compute pricing

Estimated monthly bill on Edge

Cache and queue with a ~3 GB working set, run as primary, replica and Sentinel

USD
  • Primary VM$19.24Standard: 2 vCPU, 4 GiB, 80 GB NVMe
  • Replica VM$19.24Standard: 2 vCPU, 4 GiB, 80 GB NVMe
  • Sentinel$0.00Three processes on VMs you already run
  • Storage$0.9065 GB of daily RDB snapshots: first 5 GB free, then 60 GB × $0.015
  • Egress$0.00
Total$39.38

FAQ

Redis on Edge, answered

Something else? Ask an engineer.
How does this compare to ElastiCache or Memorystore?
Cheaper, with no per-hour cache fee, and you keep root and every config option. The trade-off: you run persistence and Sentinel yourself, or our Expert Services team does.
Should I use Redis or Valkey?
They're interchangeable for almost every workload. Redis 7.4 moved to source-available licences (Redis 8 added AGPLv3 as an option); Valkey is the BSD-licensed fork of 7.2. Choose Valkey if that matters to you.
How should I handle persistence?
AOF with everysec fsync for durability, plus RDB snapshots copied to Edge Storage. For a pure cache you can turn persistence off and treat the VM as disposable.
How do I scale beyond one VM?
Add replicas for read scaling, or move to Redis Cluster across several primaries when writes or memory outgrow one VM. Both run over the private network.
How do my apps connect?
Set REDIS_URL to redis://:<password>@10.185.0.4:6379 from any VM on the same private network. Standard client libraries work unchanged; point Sentinel-aware clients at the sentinels instead.

Drop-in services

Add these without touching the stack

Shield is one server-side check and Assist is one script tag. Both work alongside whatever you already run.

Databases

Other stacks on Edge

All 38 stacks

Replace your ElastiCache node

Start free. Migrate from ElastiCache or stand up a fresh cache in a couple of minutes.

Deploy a VMPlan an ElastiCache move with us

Free tiers hard-cap. Nothing bills until you add a card.