Serverless Fully managed Apache Kafka. Nothing to size, nothing to operate.
Managed Kafka as a service, with no public internet fees
ntail is fully managed Kafka as a service: we run the brokers, the upgrades, the storage and the capacity, on infrastructure we operate ourselves in the EU and the US. You keep the Kafka clients, tools and topics you already have, pay by the hour for what you move and keep, and never pay extra for public internet traffic.
- 99.9% uptime SLA
- No maintenance windows
- EU or US region
- Free plan, no card
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, measured after compression. No cluster fee, no capacity units, no bandwidth line. Model your own workload.
What managed means, in numbers
- 99.9%Uptime SLA, in writing, with service credits
- ZeroMaintenance windows. Upgrades and scaling with no downtime for you
- <60 sFor a capacity change to land. Nothing to size up front
- $0Extra fees for public internet traffic, at any volume
Fully managed
What's managed, and what's still yours
"Managed" means different things at different providers. This is where the line sits on ntail: everything under the topic is ours to run, everything in the topic is yours.
| ntail runs | You own | |
|---|---|---|
| The cluster | Brokers provisioned, operated, upgraded and patched, with no maintenance windows | Creating a cluster in the console or the API, and picking EU or US |
| Capacity | Nothing to size or reserve. A capacity change lands in under a minute; higher limits on request | Telling us when a launch needs more |
| Storage and durability | Durable, verified storage with 11 nines of durability, sized to what you keep | Retention per topic |
| Security | TLS on every connection, SASL credentials, encryption at rest | Principals, credentials and access rules for your own teams and services |
| Visibility | Throughput and consumer lag per cluster, in the console and through the API | Deciding what to watch and what to do about it |
| The bill | Metered per started hour on compressed bytes, invoiced monthly, no hidden fees | Reading four lines |
Three ways to run it
Managed Kafka at the size you are today
Every option runs the same platform with the same clients. Start on one, move to the next without moving your data.
-
Free plan
A real managed cluster at $0 a month: 200 KiB/s in, 400 KiB/s out, measured after compression, 1-day retention and up to 10 topics. About 1,000 events a second. No card, no time limit.
-
Multi-tenant, serverless
The default. Pay as you go for data in, data out and storage from the first byte, starting from $50 a month per cluster, billed per started hour. Nothing to size, and rates that fall as you grow.
-
Dedicated cluster
Your own cluster on infrastructure we operate, fully off the public cloud if you need it, with custom data residency and volume pricing. For regulated teams and the largest workloads.
How it compares
Not every managed Kafka service is managed the same way
The brokers are somebody else's problem everywhere. What differs is what you still have to size, what the invoice counts, and where the data sits.
| ntail | Typical managed Kafka service | |
|---|---|---|
| What you size | Nothing. Create a cluster and send data | A cluster tier, a broker count or a capacity unit, chosen up front |
| What you pay for | Data in, data out and storage, measured after compression | The capacity you reserved, used or not, plus per-partition and per-hour extras |
| Public internet traffic | $0, at any volume | Metered per GB, often the largest surprise on the bill |
| Minimum commitment | $50 a month per cluster, cancel any hour | Hundreds a month, sometimes an annual contract |
| Free option | A free plan with no time limit | A trial, credits, or a tier that powers off when idle |
| Where it runs | EU or US, on infrastructure we operate ourselves; dedicated clusters fully off the public cloud | A hyperscaler region, with the hyperscaler's data transfer fees passed on |
For named vendors, modelled from their published list prices on the same workload, see ntail vs. alternatives: Confluent Cloud, AWS MSK, WarpStream, Redpanda Serverless and self-managed Kafka.
What teams run on it
One managed Kafka, four jobs
-
The event bus behind your services
Durable, always-on messaging between microservices, with one producer and as many readers as you like and no per-destination fee. Event-driven apps
-
Real-time analytics pipelines
Feed Spark, Flink, ClickHouse and your warehouse from one stream, with capacity that follows a peak and comes back down the same evening. Real-time analytics
-
Logs, metrics and traces
A buffer in front of your observability stack that compresses well, so the bytes you pay for are far fewer than the bytes you log. Observability pipelines
-
The same workload, for less
Already on a managed Kafka service? Send us last month's invoice and we'll model it on ntail, line by line. Cut streaming costs
FAQ
Managed Kafka, answered
What is managed Kafka?
Managed Kafka is Apache Kafka run for you by a provider: the brokers, upgrades, patching, storage, failover and security are the provider's job, and you connect your producers and consumers to a cluster you did not have to build. ntail is managed Kafka that is also serverless, so there is no broker count or instance size to choose either.
What is Kafka as a service, and is it the same as managed Kafka?
Yes, the two names describe the same thing: Kafka delivered as a service you subscribe to, rather than software you install and operate. The difference between providers is in what you still have to size and what you pay for. On ntail you size nothing and pay by the hour for data in, data out and storage, with no extra fee for public internet traffic.
What is the difference between managed Kafka and serverless Kafka?
Most managed Kafka services still make you pick a cluster size, a broker count or a capacity unit, and bill you for it whether or not you use it. Serverless Kafka removes that step: you create a cluster, send data, and pay for the data you move and keep. ntail is serverless. A capacity change lands in under a minute, idle capacity costs nothing, and higher limits are available on request.
Is ntail real Apache Kafka? Will my existing clients work?
Yes. Every Kafka client, framework, connector and tool you use today connects to an ntail cluster as it connects to any Kafka cluster: a bootstrap address, TLS and SASL credentials. Moving over is a connection string and a credential, not a rewrite.
What do I still have to manage?
Your side of Kafka: topics and their retention, producers and consumers, consumer groups, and who may read and write what. ntail runs everything underneath, including the brokers, upgrades, capacity, storage and encryption, and shows you throughput and consumer lag in the console and the API.
How is managed Kafka priced on ntail?
Three line items: data in and data out at $25 per 1 MiB/s a month each, measured after compression, and storage at $20 per TB retained a month. Public internet traffic is $0 at any volume. Everything is billed per started hour and invoiced monthly, with a minimum of 2 MiB/s of combined data transfer per cluster, which is $50 a month. A workload of 10 MiB/s in, 20 MiB/s out and 7 days of retention comes to about $876 a month, and the rates fall as you scale.
Where does my data live?
In the region you choose when you create a cluster, EU or US. ntail is operated by a company registered in Sweden, on infrastructure we operate ourselves. Teams with stricter requirements can get a dedicated cluster, fully off the public cloud.
Is there a free plan?
Yes. The free plan is a real managed Kafka cluster at $0 a month, with no card and no time limit: 200 KiB/s of data in and 400 KiB/s of data out, measured after compression, 1-day retention and up to 10 topics. That is about 1,000 events a second of typical JSON.
How do I move from my current Kafka or managed Kafka service?
Create a cluster, recreate your topics, then point producers and consumers at the new bootstrap address with the credentials from the console. Your clients and tooling do not change. If you want to see the cost first, send us last month's streaming invoice and we will model the same workload on ntail, line by line, with no obligation.
Managed Kafka you don't have to think about
Start on the free plan today — no card, no time limit — or send us last month's streaming invoice and we'll model the same workload on ntail, line by line, no obligation.