Status & SLA
Uptime targets by tier
| Tier | Monthly uptime target | Service credit on miss |
|---|---|---|
| Sandbox | best-effort | none |
| Starter | 99.5 % | 10 % monthly fee per 0.5 % under target |
| Growth | 99.9 % | 25 % monthly fee per 0.1 % under target |
| Scale | 99.95 % (custom) | per signed SLA |
Uptime is measured per-endpoint at the public CDN edge by an independent prober (UptimeRobot, sampling every 60 s from Mumbai + Singapore + Frankfurt vantage points). Excluded from uptime: planned-maintenance windows announced ≥7 days in advance, force-majeure (e.g., Cloudflare global outage), and customer-side rate-limit-induced 429s.
Latency targets (steady-state)
| Operation | p50 | p95 | p99 |
|---|---|---|---|
| Single-record GET | 80 ms | 250 ms | 800 ms |
| List GET (≤ 25 records) | 120 ms | 400 ms | 1,200 ms |
| Webhook delivery | 2 s | 8 s | 30 s |
Latency excludes time-to-first-byte from the upstream regulator (live-tier endpoints), which can occasionally exceed 5 seconds during regulator portal congestion.
Planned maintenance
- Window: Tuesday 02:00–04:00 IST (UTC+5:30)
- Frequency: ≤ 2 maintenance windows per quarter, no more than 1 per month
- Notice: ≥ 7 calendar days in advance via email to all paid-tier tech contacts
We aim for zero-impact maintenance (rolling deploys behind the load balancer). When a window is impactful, the announcement specifies which endpoints are affected and the expected outage duration.
Incident communication
| Severity | Definition | Initial customer notification | Update cadence |
|---|---|---|---|
| SEV-1 | Multi-tenant outage, data integrity at risk | within 30 minutes | every 30 minutes |
| SEV-2 | Single-tenant or single-endpoint outage | within 1 hour | every 1 hour |
| SEV-3 | Degraded performance, partial functionality | within 4 hours | every 4 hours |
| SEV-4 | Cosmetic / non-customer-impacting | next business day | end-of-incident only |
All incidents are followed by a postmortem within 5 business days. Postmortems are published (anonymized) at sahayakonline.co.in/incidents — currently empty (no public incidents to date — the page will populate as we operate at scale).
Live status page
We are migrating to a public status page in Q3 FY26. Current operational status is available on request from prafful@sahayakonline.co.in.
Deprecation policy
- Endpoint deprecation: 12-month notice via email. The old endpoint serves a
DeprecationHTTP header and aLinkheader pointing to the replacement. - Field deprecation within an endpoint: 6-month notice. Both fields populated during the overlap window.
- Schema-breaking changes: never within a major version. Major version bumps (v1 → v2) are co-deployed; v1 sunsets 12 months after v2 stable release.
Support response times
| Tier | Channel | Response SLA |
|---|---|---|
| Sandbox | 48 hours | |
| Starter | 24 hours | |
| Growth | Email + Slack | 4 hours, Mon–Fri 09:00–19:00 IST |
| Scale | Email + Slack + Phone | 1 hour, 24×7 |
Founder-direct support means an actual engineer who has read the code paths replies — not a tier-1 ticket-classifier.