Serverless No brokers to run, no capacity to reserve.
Serverless Kafka as the event bus behind your services
Durable, always-on messaging between your microservices — with a 99.9% uptime SLA and no brokers, no capacity planning, and no on-call rotation for Kafka itself.
- No brokers to size
- 99.9% uptime SLA
- EU data residency
- Free migration
Example monthly invoice
10 MiB/s in · 20 MiB/s out · 7-day retention- Data in10 MiB/s × $25
- $250
- Data out20 MiB/s × $25
- $500
- Storage7 days of writes ≈ 6.3 TB × $20
- $126
- Public internet trafficData sent and read, any volume
- $0
Illustrative example at list price — one producer service, two consuming services reading the same stream independently. Model your own workload.
- 99.9%Contractual uptime SLA
- ZeroMaintenance windows
- $0Extra fees for public internet traffic
- 11 9sData durability
Why teams move
Why teams stop self-managing their event bus
Kafka disruptions become someone's 3am page. We carry the operational burden and put it behind an SLA with service credits.
-
No on-call rotation for Kafka itself
No brokers to count, no partitions to plan, no upgrades to schedule. There is nothing for you to provision and we carry the pager for the platform.
-
No broker sizing before a launch
There are no brokers to size and no capacity to reserve, so a new service going live doesn't need a sizing exercise ahead of it.
-
Predictable, per-hour billing
Billed for what you actually use, per started hour — a quiet week costs less than a busy one, with no reserved capacity bought up front and no annual commitment.
What you get
What your services get
One durable log, standing behind however many services need to read from it.
-
One durable log, many readers
Order processing, notifications and audit logging can all read the same stream independently — as separate consumer groups, at no extra cost per consumer.
11 9s durability
-
Works with your client libraries
Kafka-compatible with the tools your team already uses — no proprietary client to adopt in your services.
Free migration
-
An SLA your services can depend on
A contractual 99.9% uptime SLA with service credits, not a marketing number — clusters span independent failure domains, and a lost node is replaced in seconds.
Under 44 min downtime / month
Reliability & performance
The numbers we sign up to
With service credits attached when we miss them.
- 99.9%Uptime SLAUnder 44 minutes a month, with service credits
- ZeroMaintenance windowsUpgrades and scaling with no downtime for you
- 11 9sData durabilityEvery write on durable, verified storage
- <60 sCapacity changesScale up for a peak, down the same evening
FAQ
Common questions about event-driven apps
Do I need to run or size brokers myself?
No — it's fully serverless. There's nothing to size and no capacity to reserve up front. You create a cluster, get an endpoint, and start sending; each cluster carries a throughput limit we raise on request.
What happens if there's an outage?
Production clusters carry a contractual 99.9% monthly uptime SLA with service credits if we miss it. See security for how we run the underlying infrastructure, and our status page for live status.
Can different services consume the same events independently?
Yes. Multiple consumer groups can read the same topic independently — order processing, notifications and audit logging can all subscribe to the same stream — and reading data out is never billed extra for public internet traffic, regardless of how many services are consuming.
Is our event data stored in the EU?
Yes, by default. Data is processed and stored in the EU; other regions, including the US, are available on request. See security for the full detail.
Can I start on the free plan?
Yes. The free plan is a real managed cluster at $0 a month, with no card and no time limit: 200 KiB/s in and 400 KiB/s out, measured after compression, which is about 1,000 typical JSON events a second, enough for most services' production traffic. Upgrade to the paid plan when you outgrow it and your clusters stay where they are.
Give your services an event bus you don't have to run
Start on the free plan — about 1,000 events a second, no card, no time limit — and upgrade only when your services outgrow it.