A year of hosting in Palitana: a retrospective

We’ve been hosting customers from the Palitana data center for a full year now. Here is the retrospective — what worked, what didn’t, and what we’re planning for year 2.

The numbers

We held off publishing exact figures in the launch post. What we can share without risking drift:

• Latency. Median inside-India latency is delivering low latency for Indian visitors. Anything faster for an Indian-visitor workload is now bottlenecked on the ISP edge, not on us.
• Uptime. Tied to the live status feed. Publish there, not on the blog.
• Workload mix. A mix of WordPress, ERPNext, and one-off apps. We share our own internal staging stack shape on / if you want a feel for the deployment pattern.
• Support hours. IST, with a small US-shift overlap for the European customer hours. WhatsApp first; email fallback when WhatsApp is unreachable.

The outages (retrospective)

These three are the postmortems from the build-up to launch, not live incident records - the operational feed at /status is the canonical source of live service state. We share them here because they shaped how we think about operational risk:

1. Generator transfer-switch failure (build-up). The UPS bridge worked for 30 minutes as designed; the generator had a faulty transfer switch. Lesson: test the generator transfer switch quarterly, not annually. We now run a quarterly drill.
2. CDN cache-key derivation bug (build-up). Static assets shared a key across legitimately different URLs. Lesson: derive cache keys from the full URL, not just the path.
3. Database connection-pool under-sizing (build-up). The pool was hard-coded to 10 connections; load-testing had shown we needed 40. Lesson: load-test every tier quarterly, not just at launch.

The full live incident record (when we have one) ships at /status/incidents.json.

The wins

• Latency. Hosting in India means our customers’ visitors in India get lower round-trip times than they would from a US-hosted origin. Most of our customers’ customers are in India, so this matters.
• WhatsApp support. A real engineer reads your message inside the same shift. No ticket portal, no email-only support, just WhatsApp. We don’t quote a published response-time SLA in this blog post because we want the /status + /contact surfaces to be the canonical source of truth.
• INR pricing. The plan-page INR price is the invoice price. No FX surprises. GST-compliant invoices on request.

The lessons

1. Document the failover matrix. When the power went out, the runbook said “transfer to generator after UPS bridge” but didn’t cover what to do if the transfer switch failed. We now have a 2-page decision tree + a quarterly drill.
2. Invest in operational observability. The CDN cache poisoning took 4 hours to diagnose because we didn’t have the right dashboards. We now have a Grafana + Prometheus stack on the same network with 30-second resolution metrics + PagerDuty alerting.
3. Capacity-planning is ongoing. The database pool saturation taught us that load-testing at launch isn’t enough. We now re-run quarterly.

The roadmap for year 2

• Multi-region replication. Currently single-region (Palitana only). Adding a second region (Bangalore) for redundancy is a Q1 priority.
• Edge compute. Custom routing logic at the CDN edge for traffic shaping. Not a CDN-managed feature; we’d ship a small Node service.
• GPU plans. A handful of customers have asked for GPU-backed containers for ML inference. The L3+ tier on the dedicated pool has the PCIe lanes for it. We publish the GPU plan SKU and pricing in /pricing when it ships; for early access, message us on WhatsApp.
• Backup target upgrade. Currently the daily backups land on the S3 Standard tier. We’re moving them to the new Infrequent Access tier to save cost on the 7-day retention that’s not in the product yet but is on the roadmap.