Multi-cloud · failover

Your app shouldn't die because AWS us-east-1 is having a bad day.

Cendriix orchestrates across AWS, GCP, Azure, and OCI. Pin to one cloud, span two, or let Cendriix decide based on live cost and uptime data. We handle DNS, secrets, and data residency for you. Your app stays up when a region doesn't.

Book a pilot Book a multi-cloud demo

Four failover modes.

Every mode is configured via one policy file. No bespoke runbooks. No on-call engineering in the failover path, Cendriix executes it.

PinRecommended early-stage

Single cloud · single region

One cloud. One region. Simplest topology. Recommended for early-stage products that want maximum development velocity and minimum operational surface area.

Active-passive

Two clouds · ~90s RTO

Two clouds. Cendriix continuously replicates state to the passive side. If the active region goes down, Cendriix promotes passive and re-routes DNS in under 90 seconds, no human needed.

Active-active

Two clouds · sub-5s RTO

Both clouds serve live traffic simultaneously. Cendriix splits traffic via weighted DNS, reconciles writes using a CRDT-backed merge layer, and keeps latency to the nearest region.

Sovereign

Customer-region pinning

EU users hit eu-west-1; US users hit us-east-1. Data residency enforced at the DNS layer. GDPR compliance maintained without bespoke engineering.

What Cendriix does automatically when a cloud goes down.

Seven steps. Zero pager-duty for your engineers at 3 AM. Every action is logged to the immutable audit trail.

Detect

Three independent health sources cross-reference: provider status page, Cendriix synthetic probes, your app-level metrics.

agent.sre.healthcheck.multi_source(region="us-east-1")

Drain

In-flight requests are allowed to complete. New connections stop routing to the failing region. Connection draining window: 30s.

agent.sre.traffic.drain(region="us-east-1", grace_s=30)

Promote passive

The standby environment is promoted to active. Read replicas become writers. Sequence counters are reconciled.

agent.sre.failover.promote(target="eu-west-1", mode="active-passive")

Re-route DNS

Route 53 / Cloud DNS / Traffic Manager health checks are updated. DNS TTL flushes in < 30 seconds. Users see no interruption.

agent.sre.dns.failover(ttl=30, provider="route53")

Notify

Slack, PagerDuty, and email alerts fire simultaneously. Severity, affected region, estimated RTO, and a link to the Cendriix ops console.

agent.notify.incident(severity="p1", channels=["slack","pagerduty","email"])

Draft postmortem

Cendriix writes a structured postmortem draft: timeline, root-cause hypothesis, action items. Placed in your incident tracker for human review.

agent.docs.postmortem.draft(incident_id="inc-2026-0519-01")

Rebalance

When the failed cloud recovers, Cendriix re-balances traffic on your configured schedule, not automatically, to avoid flapping.

agent.sre.traffic.rebalance(schedule="business_hours", confirm=true)

See this in action: open the sandboxes console

Data residency.

Choose a region at workspace creation. Orchestrator state and audit events stay in that region. Your source code, secrets, and production data never leave your cloud.

Region	Data center	GDPR	CCPA	HIPAAⁱ	Available clouds
us-east-1	N. Virginia	—	Yes	Yes	AWSCendriix
us-west-2	Oregon	—	Yes	Yes	AWSGCPCendriix
eu-west-1	Dublin	Yes	—	—	AWSCendriix
eu-central-1	Frankfurt	Yes	—	—	AWSAzureGCP
ap-southeast-1	Singapore	—	—	—	AWSAzure
ap-northeast-1	Tokyo	—	—	—	AWSGCP
gov-cloud	US GovCloud West	—	Yes	Yes	AWS GovCloud

ⁱ HIPAA BAA is on the roadmap, not yet available. No HIPAA-covered data should be processed until a BAA is executed.

Cost-aware multi-cloud.

Cendriix watches spot and on-demand prices across all provider regions, updated every five minutes. If GPU inference is 40% cheaper on Azure this hour, Cendriix can shift batch workloads there automatically , you set the policy in one YAML file. Cost savings are reflected on your next invoice. You never rewrite a line of code.

# cendriix.policy.yaml

cost_routing:

batch_inference: cheapest_region

api_serving: pinned_us_east_1

max_hourly_delta_pct: 15

Cendriix writes the workload migration. You set the cap. We never move more than the configured delta without explicit approval.

Ready to stop worrying about us-east-1?

Cendriix handles the failover. You ship the product.

Book a pilot Security No lock-in promise Book a demo