Datadog is too expensive: we cut our monitoring bill by 90%

If you're running Docker containers in production and paying for Datadog, you already know the feeling. That invoice that starts reasonable in January and triples by March. The one where you're paying per host, per custom metric, per GB of logs — and somehow still can't get a clear answer on what your actual monthly bill will be.

We built Kernus because we were tired of it. Not tired of Datadog as a product — it's genuinely good software — but tired of a pricing model that punishes you for actually using monitoring the way it's supposed to be used. This post isn't a hit piece. It's a real Datadog alternative cost analysis with actual numbers, from a team that switched and cut their monitoring bill by over 90%.

The real cost of Datadog for a 20-host team

Let's skip the marketing pages and do the math. Here's what Datadog actually costs when you're running a modest Docker setup — 20 hosts, each running 5-10 containers. This is a typical small-to-mid startup or a bootstrapped SaaS.

Infrastructure monitoring

Datadog's Infrastructure plan starts at $15/host/month (annual) or $18/host/month (monthly). For 20 hosts:

• 20 hosts × $15/mo = $300/mo (annual)
• 20 hosts × $18/mo = $360/mo (monthly)

This gets you basic metrics: CPU, memory, disk, network. Useful, but it doesn't include container-level visibility by default — you need the Containers add-on or the Pro plan.

Container monitoring

If you want to see individual container metrics (which, if you're running Docker, you absolutely do), you need the Pro plan at $23/host/month:

• 20 hosts × $23/mo = $460/mo (annual)

Now you're seeing container CPU, memory, and restarts. But custom metrics? Those are extra.

Custom metrics

Datadog gives you 100 custom metrics per host on the Pro plan. Anything beyond that is $0.05/custom metric/month. Emit 500 custom metrics across your fleet? That's an extra $20/mo. It sounds small until you realize almost every real application emits way more than 100 metrics per host.

Log management

This is where the bill explodes. Datadog's log management pricing:

• Ingestion: $0.10/GB ingested
• Retention (15 days): $1.70/million log events
• Retention (30 days): $2.50/million log events

A 20-host Docker setup with reasonably chatty containers generates 50-200 GB of logs per month. Let's say 100 GB:

• Ingestion: 100 GB × $0.10 = $10/mo
• Retention (15d): ~20M events × $1.70/M = $34/mo

That's $44/mo just for basic log retention. And if you need 30-day retention? $50-60/mo.

The real total

And that's conservative. We've talked to teams paying $2,000-5,000/mo for 20-50 hosts on Datadog. The moment you add APM, Continuous Profiler, or Real User Monitoring, the bill compounds fast.

What you actually need for Docker monitoring

Here's the thing most monitoring vendors won't tell you: if your primary workload is Docker containers, you don't need 90% of what Datadog sells.

You need:

• Container status — is it running or did it crash?
• CPU and memory — is something eating resources?
• Restart count — is a container stuck in a crash loop?
• OOM kills — did Docker kill it because it ran out of memory?
• Exit codes — why did it die? (code 137 = OOM, code 143 = SIGTERM, code 1 = app crash)
• Alerts — tell me immediately when something breaks
• History — show me what happened over the last 7-30 days

You don't need:

• 800+ integrations you'll never configure
• ML-powered anomaly detection for a 20-host fleet
• APM distributed tracing for a monolith
• A $200/mo Network Performance Monitoring add-on

Kernus was built around this realization. Focus on what Docker teams actually need. Cut everything else.

Kernus pricing: the 90% cut

Kernus has four plans. Here's what they cost:

For the same 20-host team paying $524/mo on Datadog for basic container monitoring? On Kernus, that's $99/mo on the Business plan. That's an 81% reduction. And if you only have 5 hosts? $29/mo. That's a 94% cut.

No per-metric pricing. No per-GB log fees. No invoices that surprise you.

Side-by-side comparison

The tradeoff is clear: if you need APM, distributed tracing, or monitoring for Kubernetes, Kafka, RDS, and 400 other services — Datadog is the right tool. If you need Docker container monitoring with alerts and you want to stop bleeding money — Kernus does the job at a fraction of the cost.

How Kernus keeps costs low

We didn't just slap a lower price tag on the same architecture. Kernus is fundamentally cheaper to operate because of three architectural decisions:

1. Push-based edge processing

Datadog's agent collects raw metrics and ships them to Datadog's cloud for processing. This means high bandwidth, high storage, and high processing costs — all of which get passed to you.

Kernus flips this. Our agent runs on your hosts and does 80% of data processing at the edge. It calculates aggregations, detects state changes, and only sends the metrics that matter. The result: dramatically less data hitting our servers, which means dramatically lower infrastructure costs.

2. ClickHouse for time-series storage

Most monitoring tools use InfluxDB or proprietary time-series databases. We use ClickHouse, a columnar database that achieves 10:1 compression ratios on metric data. This means storing 30 days of metrics for a Business plan org costs us $5-15/month per organization — not $50-100.

If you want the technical deep dive on this decision, read our post on ClickHouse vs InfluxDB for time-series metrics.

3. Focused scope

Datadog has to maintain 800+ integrations, an APM product, a log analytics engine, a security monitoring platform, a CI visibility tool, and more. Each of those products has its own engineering team, its own infrastructure costs, its own support burden.

Kernus does one thing: Docker container monitoring with alerts. Our scope is intentionally small. Smaller scope means fewer engineers, lower infrastructure costs, and a product that's easier to understand and use.

The setup comparison

Here's the entire Kernus setup process:

# 1. Install the agent (one command)
curl -fsSL https://kernus.app/install | sh

# 2. Start monitoring (one command)
kernus token YOUR_TOKEN --host company-backend
kernus agent start

Windows: Use WSL or Git Bash for this one-liner, or use PowerShell (see Windows install).

That's it. Two commands, under 2 minutes, and you're monitoring every container on that host. No config files, no dashboard templates to import, no exporters to configure.

Compare that to Datadog:

# 1. Add the Datadog repo
DD_API_KEY=your_key DD_SITE="datadoghq.com" bash -c "$(curl -L https://install.datadoghq.com/scripts/install_script.sh)"

# 2. Edit the agent config for Docker
sudo nano /etc/datadog-agent/datadog.yaml
# Enable docker_daemon check, set tags, configure log collection...

# 3. Restart the agent  
sudo systemctl restart datadog-agent

# 4. Go to Datadog UI, create dashboards
# 5. Set up alert monitors manually
# 6. Configure notification channels
# 7. Wait 5-10 minutes for data to flow

Datadog's setup isn't hard — they have good docs. But it's significantly more configuration, more decisions, and more time. For a team that just wants to know "are my containers running and healthy?", that's unnecessary friction.

When Datadog is still the right choice

We're not going to pretend Kernus replaces Datadog for everyone. Here's when you should stick with Datadog:

• You need APM — traces across microservices, flame graphs, service maps spanning non-Docker services
• You're running Kubernetes at scale — Datadog's k8s integration is excellent and deeply integrated
• You need 800+ integrations — monitoring AWS services, databases, message queues, and external APIs from one platform
• Your team is 50+ engineers — Datadog's collaboration features, role management, and integrations with PagerDuty/Jira become more valuable at scale
• Budget is not a concern — if you're funded and the monitoring bill is noise compared to your cloud spend, Datadog's breadth is hard to beat

For everyone else — small teams, bootstrapped startups, side projects in production, devs running 2-20 Docker hosts who want monitoring without the $500/mo bill — try Kernus.

Start free

Kernus is free for 1 host with up to 6 containers, 10 alerts per day, 5-minute metric collection, and 1-day retention. No credit card, no trial that expires, no Sales call required.

Install the agent, see your containers, set up an alert for when something breaks. If it solves your problem, upgrade. If it doesn't, you spent 2 minutes finding out.

Start monitoring for free →

---

If you're evaluating monitoring tools for Docker, you might also want to read our honest comparison of Kernus vs Datadog for small teams, or our guide on how to monitor Docker containers without Prometheus or Grafana.