A year of hosting in Palitana: a retrospective
A retrospective on the first year of hosting customers from a Tier-I facility in Palitana, Gujarat. The numbers, the outages, the lessons, and the roadmap ahead.
We have run Palitana for one year. Here is what we learned — what worked, what failed, what is next.
The numbers
We kept quiet at launch. Here is what we share now.
- Latency. Median India latency is 25ms. Visitors in India load pages under 25ms. Anything faster is ISP-limited. Not our issue.
- Uptime. See /status for the live feed. That is the only source.
- Workload mix. WordPress, ERPNext, and one-off apps. See / for how we deploy.
- Support. India time, with a small US shift for Europe. WhatsApp first. Email when WhatsApp is down.
The outages
Three incidents shaped how we handle risk. These are postmortems. /status has the live feed.
- Generator switch failed. UPS bridge held 30 minutes. Generator did not start. We now test the switch every quarter.
- Cache key bug. Static files shared one key across URLs. Files overwrote each other. We now use the full URL.
- Database pool too small. Pool was set to 10. We needed 40. We now load-test every quarter.
Live records are at /security/incidents.json.
The wins
- Latency. Visitors in India get faster load times. Most of our customers’ users are in India. This matters.
- WhatsApp support. A real engineer reads your message in the same shift. No portal. No email queue. We do not publish a response SLA here. /status and /contact have the real numbers.
- INR pricing. The plan price is the invoice price. No FX surprises. GST invoices on request.
The lessons
- Document failover. The runbook said “transfer to generator after UPS bridge.” It did not cover a failed switch. We now have a 2-page decision tree. We drill every quarter.
- Monitor everything. The cache bug took 4 hours to find. We had no dashboards. We now run monitoring on the same network. Alerts fire in 30 seconds.
- Plan ahead always. Launch-day load tests are not enough. We re-run them every quarter.
The roadmap for year 2
- Second region. Currently Palitana only. Bangalore is Q1 priority.
- Edge compute. Custom routing at the CDN edge. We ship a small Node service. Not a CDN-managed feature.
- GPU plans. Some customers asked for GPU-backed containers. The top dedicated plan has the slots. We publish price in /pricing when ready. Message us on WhatsApp for early access.
- Cheaper backups. Daily backups sit on S3 Standard. We move them to S3 Infrequent Access. This saves cost. It pairs with the 7-day retention upgrade.