Veritas
autonomous health
for distributed databases.
Watch Veritas heal a split-brain
A 30-second capture of a real split-brain scenario — detection, quarantine, and recovery, all without a human in the loop.
Veritas detects and recovers from a split-brain in under 30 seconds.
See it heal, live
The panel below polls the cluster every three seconds. Hit Trigger simulated failureand watch replica-1 drop offline, the orchestrator promote replica-2, and the cluster restore itself — in roughly thirty seconds.
Live cluster
3-node Postgres cluster
Trigger a failure
Inject a replica-1 outage
Watch the orchestrator detect lag, classify the replica as lag_exceeded, promote replica-2, and bring the cluster back to healthy in roughly 30 seconds.
Event log
- awaiting trigger…
What just happened
Veritas detected replica-1’s replay lag, classified it lag_exceeded, then promoted replica-2. Replica-1 rejoined as a streaming replica with minimum priority — the documented caveat from scripts/remediate.ts.
From detection to recovery in one loop
Veritas runs a continuous observe → diagnose → act → verify loop, completely independently of your team.
Engineered for the failure modes that matter
General-purpose monitoring tools treat distributed databases like monoliths. Veritas understands consensus protocols, split-brain scenarios, and the subtle ways replicated systems degrade — and knows how to bring them back.
- Purpose-built for CockroachDB, PostgreSQL, Cassandra, and YugabyteDB
- SOC 2 Type II audit trail on every action taken
- Single-digit-second MTTR across all monitored clusters
- API-first integration with your existing on-call stack
Detection stops at the alert. Veritas keeps going.
Datadog and Metaplane tell you something is broken. Oracle Autonomous fixes only Oracle. Veritas detects, diagnoses, and heals the whole cluster — including Postgres, CockroachDB, Cassandra, and YugabyteDB — in seconds.
| Capability | Traditional monitoringDatadog / Metaplane style | Managed DB autopilotOracle Autonomous / RDS | VeritasMulti-engine self-healing |
|---|---|---|---|
| Node failure detection | |||
| Root cause identification | in preview | ||
| Automated remediation | in preview in preview | ||
| Replication recovery | in preview in preview | ||
| Human on-call required | |||
| Time to full recovery (MTTH) | in preview median TBD, v0.2 |
Veritas: in preview
Veritas: in preview
Veritas: in preview
Veritas: median TBD, v0.2
Benchmark methodology
Measured on a 3-node self-hosted Postgres cluster, single-node kill test, N=10 trials, median [TBD] seconds. Number lands with v0.2 alongside the first real-cluster rollout.
Methodology script in scripts/benchmark-mtth.ts (follow-up, requires the docker-compose cluster to actually run against a reachable DATABASE_URL).
Become a design partner — measure Veritas against your own MTTH
We'll run the same single-node-kill benchmark on your cluster and publish both numbers side by side.
Contact us at veritas-2-2@polsia.app
Transparent cost. No sales call to read a number.
Veritas prices on healed nodes and prevented incidents— not on raw node count or per-seat monitoring licenses. Pick the tier that matches your workload; every plan shares the same remediation engine.
- Full autonomous healing on local and single-node dev clusters
- Unlimited simulated failures — chaos-inject any node
- Sandbox orchestration with the same remediation engine as production
- Community support and public incident-replay reports
- Unlimited healed nodes and prevented incidents per cluster
- Live production clusters: CockroachDB, PostgreSQL, Cassandra, YugabyteDB
- Sub-30-second detection-to-recovery MTTR with verification step
- API-first integration with PagerDuty, Slack, and your on-call stack
- Dedicated orchestrator tier with custom consensus-protocol remediation
- SOC 2 Type II audit trail on every action taken
- Air-gapped deployment, private CA, and regional data residency
- Named on-call engineer with quarterly resilience reviews
Prefer email? veritas-2-2@polsia.app
Pay only when Veritas actually heals something.
Pro meters node-hours under active healing — the time the orchestrator is actively running remediation on a node, not the time the node is just being observed. Quiet clusters stay quiet on your invoice. Nodes that never need a fix pay $0.
Rounding is per-minute. Every charge is line-itemed against the incident that caused it, so your bill reads like an incident report, not a usage chart.
Same cluster. Half the line item.
Datadog charges per host, around the clock — whether it ever raises an alert or not. Veritas charges per shard, only for the node-hours it actually heals. Here is what that looks like at three common cluster sizes.
| Cluster | Shards | Veritas ProPer shard · heal-only | Datadog Infrastructure ProPer host · always-on | Net savings |
|---|---|---|---|---|
| Small | 10 | ~$20 / mo*$0.40 / node-healed hr · ~5 hr heal/mo each | $150 / mo10 hosts × $15 / mo | 87% less |
| Medium | 100 | ~$200 / mo*$0.40 / node-healed hr · ~5 hr heal/mo each | $1,500 / mo100 hosts × $15 / mo | 87% less |
| Large | 1,000 | ~$2,000 / mo*$0.40 / node-healed hr · ~5 hr heal/mo each | $15,000 / mo1,000 hosts × $15 / mo | 87% less |
Methodology & assumptions
* Veritas Pro columns are illustrative. They are not a published Veritas rate card — they show a modeled cost for the Pro plan at an assumed heal-utilization, against Datadog's publicly listed per-host number, so the two are directly comparable.
- Veritas mode: Pro plan, billed at $0.40 per node-healed hour.
- Heal utilization: ~5 node-hours of active healing per shard per calendar month (illustrative).
- Datadog column: publicly listed Infrastructure Pro at datadoghq.com/pricing ($15 / host / month, one host per shard).
- Datadog pricing excludes APM, logs, custom metrics, and any add-on SKUs — those are billed separately and push the Datadog total higher in real invoices.
Compare Veritas against your own Datadog invoice
Hand us a Datadog line item and your shard count — we'll model the side-by-side against your cluster and send back the comparison in under a day.
Contact us at veritas-2-2@polsia.app
Database reliability questions, answered
The questions DBREs ask us most often before handing autonomous healing the keys to a production cluster.
Ready to eliminate unplanned downtime?
Join the teams running critical database infrastructure with autonomous self-healing powered by Veritas.
Contact us at veritas-2-2@polsia.app