Datadog's pricing is famously brutal. A single HN thread, now passed around like folklore, documented a $65 million annual bill. The secret is in the math: per-host fees, per-GB ingestion, per-custom-metric surcharges, and per-user seats stacked on top of each other. Monitor 100 hosts and the invoice mutates into something no one predicted in the original purchase order.
One of them ran a startup in Bengaluru. No observability tooling, because Datadog was unaffordable. APIs breaking. Payments failing. A team in war-room mode with nothing but logs. He knew exactly what the problem was. He just couldn't pay to solve it.
Founded by Pranay Prateek (CEO) and Ankit Nayan (CTO).
This Week's Breakdown: SigNoz
What they do: SigNoz is an open-source observability platform that puts metrics, logs, and traces in a single application. Built natively on OpenTelemetry, it runs on your infrastructure or SigNoz's managed cloud. No per-host fees. No per-seat fees. No custom-metric surcharges for using open standards. Pay for what you ingest, or self-host for free.
The numbers: 27,000+ GitHub stars. 200+ contributors. 6,000+ Slack community members. 10M+ Docker pulls since launch. 254 GitHub releases (May 2026). ~50 employees. $6.5M raised total (YC W21, SignalFire, Uncorrelated Ventures, with GitHub co-founder Tom Preston-Werner as an angel). Cloud from $49/month, enterprise from $4,000/month. ~$3M annual revenue (LinkedIn estimate, not officially confirmed). Teams in India and the US. Users in 30+ countries.
How They Built It
| Layer | Choice | Why |
|---|---|---|
| Frontend | React + TypeScript (Redux) | Single-page dashboards with virtualisation and progressive loading for real-time trace/log views |
| Backend | Go | Query service, OTel collector, alert manager. Go's concurrency model handles high-throughput telemetry ingestion |
| Database | ClickHouse | Columnar OLAP store; 8.5x lower memory than prior Kafka+Druid stack; handles high-cardinality data without limits |
| Ingestion | SigNoz OpenTelemetry Collector (Go) | Accepts any OTLP data, writes directly to ClickHouse; no Kafka dependency at standard scale |
| Auth | SSO + SAML (cloud/enterprise tiers) | SSO/SAML gated to paid tiers; community edition handles its own |
| Payments | Stripe (inferred) | Cloud subscription billing |
| Hosting/Infra | Docker, Kubernetes/Helm, AWS/GCP/Azure | Self-host via Docker Compose or Helm; BYOC for enterprise; managed cloud option |
| Storage | ClickHouse + S3 cold storage | Retention tiers: 15 days to 1 year in ClickHouse; cold logs pushed to object storage |
| API/Integrations | PromQL + ClickHouse SQL + OpenAPI; Slack, PagerDuty, OpsGenie, MS Teams; AWS/Azure/GCP native integrations | Query builder covers most use cases; PromQL for teams migrating from Prometheus |
Stack from GitHub repo (TypeScript 52.9%, Go 38.4%), job postings, single-binary architecture release notes (v0.76.0), and SigNoz engineering blog.
The Real Story
The name comes from electrical engineering. Pranay and Ankit are both EE graduates (Pranay from IIT Madras, Ankit from IIT Delhi). Signal versus noise. The problem of finding what matters inside the chaos of data was the literal subject of their degrees before it became the name of their company.
Pranay worked at Microsoft in the CRM Dynamics team after college, then came back to Bengaluru to lead product at a startup. That's where he first felt the problem. The company used Elasticsearch for logs but couldn't find people who knew how to run it. Prometheus and Grafana sat unconfigured. When something broke in production, no one had a trace. Ankit was leading engineering teams in similar environments, watching the same pattern: microservices creating cascading failures that took hours to debug because no one had instrumented anything properly.
They started building in October 2020. The initial architecture was ambitious: Kafka for ingestion, Apache Druid for storage. Both are legitimate enterprise-grade systems. Druid powers Lyft and Airbnb at scale.
They launched on Hacker News in February 2021. The feedback was swift. Multiple engineers pointed out that the Kafka+Druid combo was too heavy to run locally. One comment stuck: "If step one of your setup instructions is 'provision a Kafka cluster,' you are going to lose 90% of people right there."
By mid-2021, they had migrated to ClickHouse. The startup memory footprint dropped 8.5x. People could try it on a laptop. Stars accelerated. Community members who'd bounced off the first version came back. Contributors started appearing. Customers who became contributors appeared after that.
The cloud product was an accident. The team went on a workation in Goa in December 2023 and decided to do a one-week soft launch of SigNoz Cloud to test for issues, then roll it back. They got four signups. They never rolled it back.
In 2024, they consolidated the query service, frontend, and alert manager into a single binary. Three components instead of five. The thesis was the same as the Druid migration: remove friction. If people can't install it easily, they won't. The product earned its complexity only after it earned trust.
The Marketing Playbook: Content as Community
SigNoz doesn't run paid ads. Growth is almost entirely content and community.
The content strategy is built on a single observation: when multiple people get stuck on the same thing in Slack or GitHub, that's a gap on the internet that needs filling. The SigNoz team writes the definitive resource on that topic. OpenTelemetry was a young project when they started; most content explaining it in practical terms didn't exist. SigNoz filled that vacuum. They now rank for hundreds of long-tail observability and OTel keywords.
The Slack community is 6,000+ members. Monthly "SigNal" community newsletters publish product updates, contributor spotlights, and industry news. They trend on Hacker News deliberately: deep technical posts on Kafka architecture and OpenTelemetry internals get to the front page, then developers visit the repo.
There's a conversion flywheel underneath all of this. A developer googles a ClickHouse or OpenTelemetry question, finds a SigNoz blog post, installs the open-source version, self-hosts it for months, and eventually upgrades to cloud because the ops overhead of running ClickHouse themselves becomes annoying. The content seeds the install. The install seeds the upgrade.
They also publish direct comparison pages: SigNoz vs Datadog, vs Grafana, vs New Relic, vs Dynatrace. Each page targets buyers in evaluation mode. And since late 2023, an automated Datadog dashboard migration tool lets engineers move their existing dashboards to SigNoz without rebuilding from scratch. The tool converts a migration objection into a feature.
How SigNoz Makes Money
| Plan | Price | What's Included |
|---|---|---|
| Community Edition | Free (self-hosted) | Full platform; you run your own ClickHouse |
| Startup Program | $19/month | Cloud; <3 years old, <30 employees, <$6M raised |
| Teams Cloud | $49/month base | Includes $49 usage; then $0.30/GB logs+traces, $0.10/million metric samples; unlimited users and hosts |
| Enterprise Cloud | From $4,000/month | Dedicated cloud environment; HIPAA, SSO, SAML, volume contracts, SLAs |
| Enterprise BYOC/Self-Hosted | Custom | SigNoz deploys in your cloud; you own the infra, they run it |
No per-seat fees. No per-host fees. No premium charge for using OpenTelemetry standards.
The Global Landscape: Observability
The observability and APM market sits at roughly $28B in 2025, growing at 19% CAGR according to Market Research Future. Cloud-native infrastructure and AI-era complexity are the drivers: more services, more dependencies, more things to instrument.
The Big Players (Enterprise Scale)
| Company | Where | Scale | What Sets Them Apart |
|---|---|---|---|
| Datadog | US | $3.43B FY2025 revenue, ~$80B market cap, 47K+ customers | Full-stack: APM, RUM, security, synthetics, 900+ integrations |
| Grafana Labs | US | $400M+ ARR, $6B+ valuation, 7,000+ customers | Composable LGTM stack (Loki, Tempo, Mimir); community of 25M+ users |
| Dynatrace | US | $1.5B+ ARR, public (DT) | AI-powered root-cause analysis; strong in enterprise Java and SAP environments |
| New Relic | US | Private (PE buyout 2024), 175,000+ free orgs | Generous free tier; full-platform play with recent open-source push |
| Splunk (Cisco) | US | Acquired by Cisco for $28B in 2024 | SIEM + observability combined; very large enterprise footprint |
The giants own the enterprise relationships, the integrations libraries, and the compliance certifications. Datadog's 28% YoY revenue growth in 2025 shows the market is still expanding fast, not contracting.
The Mid Players (Funded, Growing)
| Company | Where | Funding / Scale | What They Do |
|---|---|---|---|
| Honeycomb | US | ~$70M raised | High-cardinality events and query-centric observability; beloved by platform engineers |
| Chronosphere | US | Acquired by Palo Alto Networks (~$3B, Jan 2026) | Cloud-native observability; Calyptia pipeline acquisition |
| HyperDX | US | ClickHouse-backed, open-source | Frontend + backend observability with session replay |
| Coroot | US | eBPF-based, Kubernetes-native | Zero-instrumentation, auto-maps service dependencies |
The Small Players (Open Source / Indie)
| Company | Where | Funding / Scale | What They Do |
|---|---|---|---|
| SigNoz | India / US | $6.5M raised, ~27K GitHub stars | Unified OSS platform; ClickHouse backend; OTel-native |
| OpenObserve | US | OSS | Rust + Parquet on object storage; claims 140x cheaper than Elasticsearch |
| Uptrace | Remote | OSS | OTel-native, ClickHouse backend; opinionated setup |
| Coroot | US | OSS tier | Kubernetes-native, eBPF auto-discovery |
SigNoz is the most feature-complete open-source Datadog alternative in this group. The common thread for the small players: ClickHouse as the default backend, OpenTelemetry as the default ingestion format. The architecture is converging around these two choices.
How We Got Here
| Year | What Happened | Key Event |
|---|---|---|
| 2010 | Datadog founded | Per-host SaaS APM becomes the enterprise default |
| 2014 | Prometheus open-sourced | SoundCloud metric collection; CNCF adoption follows |
| 2017 | OpenTelemetry predecessors (OpenCensus, OpenTracing) | Vendors try to standardize on instrumentation |
| 2019 | OpenTelemetry merges | CNCF's second most active project after Kubernetes |
| Oct 2020 | SigNoz starts building | Pranay and Ankit begin with Kafka+Druid |
| Feb 2021 | SigNoz launches on HN | Community feedback: Druid too heavy |
| Mar 2021 | YC W21 + $1.1M seed | SignalFire, YC, angels |
| Mid 2021 | Migrate to ClickHouse | Memory drops 8.5x; adoption accelerates |
| Sep 2023 | $5.4M led by SignalFire | Total: $6.5M; cloud + enterprise push |
| Dec 2023 | Cloud soft-launch in Goa | One week test; 4 signups; never rolled back |
| 2024 | Single binary release (v0.76.0) | Query service + frontend + alertmanager in one binary |
| 2025 | Cloud at $49/month | Down from $199; Noz AI assistant + MCP server |
| 2026 | 27K GitHub stars, 254 releases | Users in 30+ countries; handles 10TB+ ingest/day |
The Pattern You Can Steal
1. Ship the ambitious architecture, then simplify until strangers will try it. SigNoz launched with Kafka and Druid because that's what scales. The HN community told them it was too heavy for local use. They migrated to ClickHouse, cut memory 8.5x, and adoption followed. Later, five components became three. The architecture is not a technical constraint. It is a first-impression problem. If your open-source product requires ten steps to install, you're writing a rejection letter to 90% of potential users.
2. Build the migration tool before the comparison page. SigNoz published "SigNoz vs Datadog" pages, but the move that converts actual paying Datadog customers is the automated dashboard migration tool. When a competitor's customer can move their existing dashboards in minutes, you've removed the largest single objection to switching. Find the migration friction in your space and eliminate it before the prospect asks about it.
3. Content that fills actual knowledge gaps compounds faster than content that repeats what everyone already wrote. SigNoz writes about OpenTelemetry implementation details that don't exist anywhere else. That content ranks because no one else wrote it, and it attracts exactly the developers who would use the product. The distribution advantage: when a technology is new, the first credible explainer owns the keyword permanently.
Quick Hits
OpenTelemetry is now the second most active project in the CNCF, behind only Kubernetes. The standardization it represents is the rising tide SigNoz is built to capture. Every new service that adopts OTel instrumentation becomes a potential SigNoz user. opentelemetry.io
Datadog's Q4 2025 results showed 28% YoY revenue growth to $3.43B. The market is not slowing. But at $65M bills for some customers, the window for cost-sensitive alternatives keeps widening. Investor release
Palo Alto Networks acquired Chronosphere for approximately $3B in January 2026, merging cloud-native observability with security. The consolidation at the top may push mid-market buyers toward independent OSS platforms rather than getting absorbed into security vendor bundles. PANW announcement
What We're Watching
AI observability as the next frontier. SigNoz already supports LLM monitoring via OpenTelemetry (Langtrace, OpenLLMetry, OpenLIT integration). AI applications produce more traces, more logs, and more unusual failure patterns than conventional services. If AI-native startups adopt OpenTelemetry by default, SigNoz is positioned early.
The BYOC model at enterprise scale. Enterprise BYOC (bring your own cloud) is the version of self-hosting that big companies actually pay for: SigNoz runs the platform, you own the infra, your data never leaves your VPC. It solves the HIPAA and GDPR objections without asking the customer to operate ClickHouse themselves. This tier is where the real revenue ceiling sits.
Community Edition as a funnel, not a ceiling. The open-source version creates a pipeline: install freely, then hit operational complexity as you scale, then upgrade to cloud. GitHub stars are a leading indicator, not a revenue metric. The conversion rate from self-hosted community install to cloud subscriber is the number that determines whether $6.5M in funding compounds into a real business.
Ship it.
— The FounderSpec Team