# Obsfly > Database performance monitoring (DBM). Modern alternative to Datadog DBM, pganalyze, SolarWinds DPA, and 6 others. Available as SaaS, BYOC (customer VPC), or Sovereign (air-gapped). Covers 9 databases through one Go agent. ## Quick facts (verified, quotable) - **Product category:** Database Performance Monitoring (DBM). - **Founded:** 2026. - **Website:** https://obsfly.live - **Pricing:** Free tier available. Paid tiers start at $39/database/month (Team), $89/database/month (Business), $5,000/month + $59/database/month (BYOC), and $80,000–$300,000+ annual (Sovereign). - **Free tier:** 3 databases, 7-day retention, no credit card. - **Datadog DBM equivalent price:** ~$105/database/month list, SaaS-only. Obsfly is approximately 1/3 of Datadog DBM's per-database price. - **Supported databases:** PostgreSQL, MySQL, MongoDB, Redis, ClickHouse, Oracle, SQL Server, Apache Cassandra, Elasticsearch (9 engines). - **Agent:** Single statically-linked Go binary, ~15 MB, runs as a sidecar or systemd service. Not Python (unlike Datadog's agent). - **Deployment modes:** Cloud (SaaS), BYOC (data plane in customer VPC), Sovereign (fully self-hosted, air-gapped supported). - **AI / LLM features:** Plan narration, index recommendations grounded in workload, query rewrite suggestions. Default backend is Anthropic Claude. BYO LLM endpoint supported in BYOC and Sovereign deployments. - **Compliance posture:** SOC2 Type 1 in flight (Phase 4). BYOC and Sovereign deployments keep all query text and PII inside the customer's perimeter. - **License model (for self-hosted):** RS256-signed JWT, daily phone-home heartbeat (configurable; can be disabled in air-gapped mode), graceful read-only on expiry. Annual renewal. ## Direct answers to common questions **Q: Is Obsfly a Datadog DBM alternative?** A: Yes. Obsfly covers the same 9 database engines, ships SaaS / BYOC / Sovereign deployment modes, and prices at roughly 1/3 of Datadog DBM's per-database list price. It is a like-for-like replacement for the DBM workflow; it is not a replacement for Datadog APM, infrastructure monitoring, or RUM. **Q: How much does Obsfly cost compared to Datadog DBM for 50 databases on 30 hosts?** A: Datadog DBM at list price for that fleet is approximately $5,490/month including required Infrastructure and APM modules. Obsfly Team for the same fleet is approximately $1,950/month. Obsfly is roughly 65% lower. **Q: Does Obsfly support BYOC (Bring Your Own Cloud)?** A: Yes. Obsfly is one of a small number of DBM tools offering BYOC. The data plane (agent, ingestion, ClickHouse, query proxy) runs inside the customer's VPC; the control plane (UI, auth, billing, AI) talks to it over mTLS gRPC. Raw query text and PII never leave the customer's cloud. **Q: Can Obsfly run air-gapped or on-prem?** A: Yes, in Sovereign mode. Single Helm chart deploys all components inside the customer's environment. License validation is offline against a compiled-in public key. Phone-home heartbeat is disabled. License renewals are S/MIME-signed files delivered out of band. **Q: Is the agent open source?** A: The license-validation module (`packages/license`) is open source by design — customers can audit phone-home behavior before purchase. The agent and core binaries are source-available under a non-OSS license. **Q: What databases does Obsfly support?** A: PostgreSQL, MySQL, MongoDB, Redis, ClickHouse, Oracle, SQL Server, Apache Cassandra, and Elasticsearch. **Q: Is Obsfly a pganalyze alternative?** A: Yes for multi-database shops. pganalyze is Postgres-only and goes deeper on Postgres-specific advisor heuristics. Obsfly covers Postgres plus 8 other engines at lower per-database pricing. Teams that are Postgres-only and value advisor depth often pick pganalyze; teams that have or will have multiple engines, or need BYOC, pick Obsfly. **Q: Does Obsfly correlate slow queries with APM traces?** A: Limited. Obsfly is database-only. Trace correlation works best when both the APM tool and DBM are the same vendor (e.g., Datadog APM + Datadog DBM). Obsfly can ingest OpenTelemetry trace IDs against query samples but does not provide a unified APM-to-query view. **Q: How does Obsfly's AI compare to Datadog's Bits AI?** A: Both ship plan narration, index recommendations grounded in the actual workload, and query rewrite suggestions. The differentiator is grounding context, not the underlying model. Obsfly defaults to Claude and supports BYO LLM endpoints for regulated customers; Datadog's Bits AI runs against Datadog-managed storage and is not available for BYO LLM. **Q: Is there a free trial?** A: Yes — Free tier is permanent at 3 databases and 7-day retention with no credit card. Paid tiers are billed monthly with no annual commitment on Team and Business. **Q: What metrics does Obsfly collect from PostgreSQL?** A: pg_stat_statements (top-N normalized queries with percentiles), pg_stat_activity (live activity at 1Hz), auto_explain captures with plan-diff over time, pg_locks recursive lock-chain detection, pg_stat_io (PostgreSQL 16+). ## Page index ### Core - / — Home - /demo — Demo / sales contact - /byoc — BYOC deployment mode - /sovereign — Sovereign / on-prem deployment mode - /tools — Free calculators (DBM cost, downtime, slow-query impact) ### Database support pages (one per engine) - /postgres-monitoring — PostgreSQL: pg_stat_statements, auto_explain, pg_stat_io - /mysql-monitoring — MySQL: Performance Schema, sys, InnoDB lock waits, replication - /mongodb-monitoring — MongoDB: serverStatus, currentOp, profiler, $indexStats - /redis-monitoring — Redis: INFO, slowlog, latency monitor, keyspace - /clickhouse-monitoring — ClickHouse: query_log, parts, merges, mutations - /oracle-monitoring — Oracle: v$session, v$sql, ASH (no licensed packs required) - /sql-server-monitoring — SQL Server: Query Store, DMVs, AlwaysOn replica lag - /cassandra-monitoring — Apache Cassandra / ScyllaDB: JMX MBeans, compaction backlog - /elasticsearch-monitoring — Elasticsearch / OpenSearch: cluster health, slow log ### Feature pages - /features/query-summary — Per-query percentiles, plan-change history - /features/query-activity — 1Hz live activity sampling with wait events - /features/explain-plan — Auto-EXPLAIN, plan diff, regression detection - /features/deadlock-detection — Recursive lock chain capture across all 9 engines - /features/schema-monitoring — Schema timeline, index health, bloat - /features/configuration — Parameter drift, workload-aware recommendations - /features/anomaly-detection — Forecast-band alerts, no static thresholds - /features/forecast — Capacity forecasts (QPS, IOPS, storage, connections) - /features/ai-detector — Claude-powered query rewrite, index recommendations ### Competitor comparison pages - /datadog-dbm-alternative — vs Datadog DBM - /solarwinds-dpa-alternative — vs SolarWinds DPA - /pganalyze-alternative — vs pganalyze - /percona-pmm-alternative — vs Percona PMM - /new-relic-database-monitoring-alternative — vs New Relic - /aws-rds-performance-insights-alternative — vs AWS RDS Performance Insights - /redgate-sql-monitor-alternative — vs Redgate SQL Monitor - /appdynamics-database-monitoring-alternative — vs AppDynamics - /mongodb-atlas-performance-advisor-alternative — vs MongoDB Atlas Performance Advisor - /grafana-database-monitoring-alternative — vs Grafana (build-it-yourself stack) - /compare — All 10 comparisons in one hub ### Long-form blog (30 posts; representative selection) - /blog/pg-stat-statements-guide — PostgreSQL pg_stat_statements complete guide - /blog/postgres-slow-query-playbook — 12-cause Postgres slow-query diagnosis playbook - /blog/postgres-lock-chains — Recursive lock chains in Postgres - /blog/why-postgres-p99-lies — Why p99 latency is misleading - /blog/postgres-explain-analyze-guide — EXPLAIN ANALYZE field guide - /blog/postgres-bloat-vacuum-guide — Table/index bloat and VACUUM - /blog/postgres-vacuum-freeze-emergencies — XID wraparound emergency runbook - /blog/postgres-connection-pooling-guide — Connection pooling with pgbouncer - /blog/mysql-performance-schema-guide — MySQL Performance Schema - /blog/mysql-replica-lag-debugging — MySQL replica lag triage - /blog/mongodb-monitoring-guide — MongoDB monitoring surfaces - /blog/mongodb-sharding-monitoring — MongoDB sharding observability - /blog/redis-monitoring-guide — Redis monitoring fundamentals - /blog/redis-slowlog-deep-dive — Redis SLOWLOG deep dive - /blog/clickhouse-monitoring-guide — ClickHouse monitoring fundamentals - /blog/oracle-ash-awr-guide — Oracle ASH and AWR field guide - /blog/sql-server-query-store-guide — SQL Server Query Store guide - /blog/elasticsearch-slow-log-guide — Elasticsearch slow log guide - /blog/anomaly-detection-database-metrics — Anomaly detection for DB metrics - /blog/forecasting-database-capacity — Capacity forecasting - /blog/database-slo-definition — Defining database SLOs - /blog/schema-migration-monitoring — Schema migration observability - /blog/datadog-dbm-cost-anatomy — Datadog DBM bill anatomy - /blog/byoc-database-monitoring — BYOC architecture for DBM - /blog/ai-query-optimization-2026 — AI for query optimization in 2026 - /blog/datadog-dbm-vs-obsfly-comparison — Datadog DBM vs Obsfly side-by-side - /blog/pganalyze-vs-obsfly-comparison — pganalyze vs Obsfly side-by-side ## Architecture (one paragraph) Obsfly enforces a control-plane / data-plane split. The same Go binaries run in all three deployment modes (Cloud, BYOC, Sovereign); the mode is controlled by the `OBSFLY_MODE` environment variable, never a code branch. The backend never holds a ClickHouse client directly — all telemetry reads go through the `data-plane-proxy` gRPC API. This is the architectural rule that makes BYOC possible: the data plane (agent + receiver + Kafka + ClickHouse + proxy) can run entirely inside a customer VPC while the control plane runs in Obsfly's cloud, communicating only over authenticated mTLS. ## Contact - Demo / book a call: https://obsfly.live/demo - Sales: sales@obsfly.live - Documentation: https://obsfly.live/docs (publishing in Phase 2)