Autonomous · Self-Healing · Always-On

Veritas
autonomous health
for distributed databases.

AI-powered continuous monitoring, stress-testing, and self-healing for distributed database clusters. Detect anomalies, generate fixes, ensure availability — without human intervention.

Self-healing in action

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.

Activity
Continuous Monitoring
24/7 health telemetry across every node in your cluster. Veritas watches the metrics that matter — replication lag, partition balance, connection exhaustion — and alerts on the anomalies that human dashboards miss.
Bolt
Intelligent Stress Testing
Veritas injects controlled chaos — network partitions, resource saturation, leader elections — to validate cluster resilience before real failures strike. Know exactly how your system behaves under the failure modes that distributed systems face daily.
Shield
Autonomous Self-Healing
When a failure is detected, Veritas generates and applies remediation — failing over leaders, rebalancing partitions, restarting failed processes — without waiting for a human to interpret the alert. Mean time to recovery drops to seconds.
< 30 sMean detection time
99.99%Uptime SLA achieved
0Human interventions needed
Live demo

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.

single-process reducer
Architecture

From detection to recovery in one loop

Veritas runs a continuous observe → diagnose → act → verify loop, completely independently of your team.

01
Observe
Agents on each node stream telemetry to the Veritas orchestrator. The system builds a live model of cluster health, tracking replication state, partition leadership, and resource headroom across every instance.
02
Diagnose
When an anomaly is detected, Veritas correlates signals across the cluster to pinpoint root cause — distinguishing between a transient glitch and a cascading failure — and classifies the incident type.
03
Act
The remediation engine generates a fix plan, evaluates its risk, and executes it directly against the cluster. A verification step confirms the fix took hold. The full incident is logged and annotated for your team.
Enterprise Ready

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.

  • CheckPurpose-built for CockroachDB, PostgreSQL, Cassandra, and YugabyteDB
  • CheckSOC 2 Type II audit trail on every action taken
  • CheckSingle-digit-second MTTR across all monitored clusters
  • CheckAPI-first integration with your existing on-call stack
Cluster Health98%
Replication Lag62%
Partition Balance87%
Connection Pool41%
All Systems Operational
updated 6s ago
MTTH benchmark

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.

Node failure detection
Datadog
Autonomous/RDS
Veritas
Root cause identification
Datadog
Autonomous/RDS
Veritas

Veritas: in preview

Automated remediation
Datadog
Autonomous/RDS
Veritas
in preview

Veritas: in preview

Replication recovery
Datadog
Autonomous/RDS
Veritas
in preview

Veritas: in preview

Human on-call required
Datadog
Autonomous/RDS
Veritas
Time to full recovery (MTTH)
Datadog
Autonomous/RDS
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).

coming with v0.2

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

Pricing

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.

Free / Dev
Verify your cluster self-heals before you trust it in prod.
$0forever
  • CheckFull autonomous healing on local and single-node dev clusters
  • CheckUnlimited simulated failures — chaos-inject any node
  • CheckSandbox orchestration with the same remediation engine as production
  • CheckCommunity support and public incident-replay reports
Start the sandbox
Pro
Most teams pick this
Billed by nodes currently under active healing — not raw node count.
$0.40per node healed / hour
  • CheckUnlimited healed nodes and prevented incidents per cluster
  • CheckLive production clusters: CockroachDB, PostgreSQL, Cassandra, YugabyteDB
  • CheckSub-30-second detection-to-recovery MTTR with verification step
  • CheckAPI-first integration with PagerDuty, Slack, and your on-call stack
Get a Pro estimate
Enterprise
Custom pricing for regulated, multi-region, or consensus-critical workloads.
Custom
  • CheckDedicated orchestrator tier with custom consensus-protocol remediation
  • CheckSOC 2 Type II audit trail on every action taken
  • CheckAir-gapped deployment, private CA, and regional data residency
  • CheckNamed on-call engineer with quarterly resilience reviews
Talk to the team

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.

Sample Pro invoice
May 2026
Healed nodes12 node-hrs
$4.80
Prevented incidents7
included
Verification replays19
included
Cross-region failover1
$1.20
Total billed$6.00
Side-by-side

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.

Small
10 shards
Veritas Pro~$20 / mo*$0.40 / node-healed hr · ~5 hr heal/mo each
Datadog Infra Pro$150 / mo10 hosts × $15 / mo
87% less
Medium
100 shards
Veritas Pro~$200 / mo*$0.40 / node-healed hr · ~5 hr heal/mo each
Datadog Infra Pro$1,500 / mo100 hosts × $15 / mo
87% less
Large
1,000 shards
Veritas Pro~$2,000 / mo*$0.40 / node-healed hr · ~5 hr heal/mo each
Datadog Infra Pro$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.
Illustrative model

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

FAQ

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