India cloud uptime and latency report 2026
Monthly uptime for April to November 2026, measured round-trip latency from our Noida datacentre to Indian cities, and NVMe storage and network benchmark results. Aggregates only. Data as of 2026-11-20.
Summary
- Average uptime across all components for completed months: 99.943% (target 99.9%).
- 11 Indian cities measured against Noida; the lowest round trip is 5.4 ms and the highest 49.4 ms.
- 4k random read on NVMe-backed block storage: 80,088 IOPS, median of three runs.
Monthly uptime by component, April to November 2026
Same figures as the public SLA page. Nov 2026 is month-to-date at 2026-11-20. Values under 99.9% are shaded.
| Component | Apr 2026 | May 2026 | Jun 2026 | Jul 2026 | Aug 2026 | Sep 2026 | Oct 2026 | Nov 2026 (MTD) | Average, full months |
|---|---|---|---|---|---|---|---|---|---|
| Noida region gateway | 100.000% | 100.000% | 99.993% | 99.981% | 99.764% | 99.975% | 99.993% | 100.000% | 99.958% |
| S3 object storage | 100.000% | 100.000% | 99.993% | 99.981% | 99.764% | 99.975% | 99.993% | 99.992% | 99.958% |
| App hosting edge | 100.000% | 100.000% | 99.993% | 99.981% | 99.764% | 99.975% | 99.993% | 99.992% | 99.958% |
| Public website | 99.997% | 99.991% | 100.000% | 99.975% | 99.968% | 99.993% | 99.621% | 100.000% | 99.935% |
| Customer panel sign-in | 99.997% | 99.991% | 100.000% | 99.975% | 99.968% | 99.686% | 99.621% | 99.989% | 99.891% |
| API | 99.997% | 99.991% | 100.000% | 99.975% | 99.968% | 99.466% | 100.000% | 99.989% | 99.914% |
| Noida network edge, DNS and mail (SMTP) | 99.997% | 99.991% | 100.000% | 99.975% | 99.968% | 99.993% | 100.000% | 100.000% | 99.989% |
| All components | 99.998% | 99.995% | 99.997% | 99.978% | 99.881% | 99.866% | 99.889% | 99.995% | 99.943% |
Measured round trip from Noida to Indian cities
ICMP ping to our Noida server from public measurement probes inside each city, 16 packets per probe, median over three rounds.
| City | Round trip to Noida | Measured on |
|---|---|---|
| Noida | 5.4 ms | 2026-10-26 |
| Delhi | 7.0 ms | 2026-11-20 |
| Jaipur | 14.8 ms | 2026-11-08 |
| Pune | 30.1 ms | 2026-10-26 |
| Mumbai | 38.5 ms | 2026-10-26 |
| Hyderabad | 38.8 ms | 2026-10-26 |
| Chennai | 41.8 ms | 2026-10-26 |
| Indore | 46.3 ms | 2026-10-26 |
| Bengaluru | 47.9 ms | 2026-11-20 |
| Ahmedabad | 48.2 ms | 2026-11-08 |
| Kolkata | 49.4 ms | 2026-10-26 |
Round trip by ISP network
Measured on 20 Nov 2026 with the public Globalping probe network. Each probe sent 16 ICMP packets to our Noida origin server (), not to a CDN edge. Only consumer and access ISP networks are included; hosting, datacentre and exchange networks are left out. 3 rounds within ~1 hour on 20 Nov 2026 (single time window), so the figures say nothing about different times of day. A network's figure is the median of its probe averages; a city's figure is the median of its networks. Probes with more than 5% packet loss are left out. The ISP names are the network operators the probes report.
| City | ISP network | ASN | Median | Fastest to slowest | Probes |
|---|---|---|---|---|---|
| Delhi | Atria Convergence Technologies | AS18209 | 5.8 ms | 4.7 to 12.3 ms | 3 |
| Delhi | Bharti Airtel Ltd., Telemedia Services | AS24560 | 8.3 ms | 6.3 to 12.5 ms | 3 |
| Bengaluru | Atria Convergence Technologies Pvt. Ltd. Broadband Internet Service Provider INDIA | AS24309 | 42.3 ms | 41.1 to 49.3 ms | 3 |
| Bengaluru | Bharti Airtel Ltd., Telemedia Services | AS24560 | 53.5 ms | 47.8 to 62.5 ms | 3 |
Not enough data to publish a figure for: Noida (no consumer ISP probe available), Gurugram (no consumer ISP probe available), Mumbai (fewer than two consumer ISPs with usable probes), Pune (no consumer ISP probe available), Ahmedabad (fewer than two consumer ISPs with usable probes), Hyderabad (no consumer ISP probe available), Chennai (fewer than two consumer ISPs with usable probes), Kolkata (fewer than two consumer ISPs with usable probes), Lucknow (no consumer ISP probe available), Jaipur (fewer than two consumer ISPs with usable probes), Indore (no consumer ISP probe available). The older single-number estimates on those city pages are labelled with their own dates.
Download the raw per-probe measurements (CSV): date, round, city, ISP, ASN, probe, min, avg, max and loss for every probe.
Time it from your own browser with the Noida latency test. Price a server with the cost and GST calculator or size one with the VPS sizing calculator.
Storage and network benchmark results
Run 2026-10-25. 4 vCPU, 8 GB RAM, Ubuntu 24.04, KVM, Noida, NVMe-backed network block storage. Median of three runs. fio 3.36, sysbench 1.0.20, iperf3 3.16. Full write-up: NVMe benchmark post.
| Test | Result | Unit | Run-to-run range |
|---|---|---|---|
| 4k random read | 80,088 | IOPS | 77,403 to 82,186 |
| 4k random write | 19,839 | IOPS | 19,561 to 20,667 |
| 4k mixed 70/30 read/write | 41,406 | IOPS | 40,807 to 42,417 |
| 1M sequential read | 4,026 | MB/s | 3,992 to 4,073 |
| 1M sequential write | 1,822 | MB/s | 1,715 to 2,002 |
| CPU steal | 0 | % | - |
| sysbench CPU, 1 thread | 514.9 | events/s | - |
| sysbench CPU, 4 threads | 2,051 | events/s | - |
| iperf3 same zone, upload | 191.1 | Mbit/s | - |
| iperf3 same zone, download | 190.9 | Mbit/s | - |
- The iperf3 figures were capped at 200 Mbit/s by the test offering, so they show the cap, not the platform limit.
- Queue-depth-1 read p99 latency stayed within 1.09x of the median.
- Port tiers on production plans: 1 Gbps; the first 1 TB per month at 1 Gbps, 1 to 2 TB at 500 Mbps, above 2 TB at 50 Mbps.
How the numbers were produced
- Uptime. Share of each calendar month (IST) that a component answered synthetic checks from several independent probes; a failure counts only when most probes agree. Published monthly on the SLA page. Source data: https://status.bookmyhost.com/sla-monthly.json.
- Latency. ICMP ping to our Noida server from public measurement probes inside each city, 16 packets per probe, median over three rounds. One median per city; city pages list the probe method in detail.
- Benchmarks. One test VM, three runs, median reported, range shown. Results depend on the plan, the neighbours and the test; treat them as indicative, not as a guarantee.
- Scope. Aggregates only. Individual incidents, customers and internal systems are not part of this report.
- Limits. The current month is incomplete. Latency is network round trip, not page-load time. The iperf3 figures show a 200 Mbit/s test cap.
Reference and download
BookMyHost Cloud (2026). India cloud uptime and latency report 2026. Published 2026-11-20, data as of 2026-11-20. https://bookmyhost.com/reports/india-cloud-uptime-latency-2026
Is this report free to reuse?
You may quote and link it with attribution. Use the cite block on the page and link to this URL.
How is monthly uptime calculated?
It is the share of the month a component answered its checks, as published on our public SLA page. A month is a calendar month in IST; the current month is month-to-date.
Why is latency one number per city?
It is the median of repeated ICMP measurements from public probes inside that city to our Noida server, on the date shown. It is a network figure, not a page-load time.
Does the report include incident details?
No. It is aggregates only: percentages, medians and benchmark results.
Run your workload where the numbers come from
NVMe servers in Noida with DDoS protection and GST invoices. Every new account starts with ₹200 of free credit.


