Among the 12,480 production websites included in this analysis, median observed uptime was 100% during valid monitoring periods in September 2026. A total of 87.5% recorded no unplanned downtime during those periods.
This report analyses anonymised uptime-monitoring data for 12,480 production websites hosted on eWebGuru infrastructure in India, from 1 to 30 September 2026. These websites are a subset of all websites eWebGuru hosts. We publish these figures so businesses can see how often downtime actually happens, what causes it, and what they can do to reduce its impact.
Period: 1- 30 September 2026
Population analysed: 12,480 monitored production websites on eWebGuru infrastructure
Review: Anil Kumar, System Engineer, eWebGuru
Published date:
This is the first edition of a monthly series.
Key takeaways
- Outages were short. The median unplanned incident lasted 22 minutes.
- Median uptime was 100%. 8,216 of 12,480 websites (65.8%) had no downtime of any kind during the month.
- 87.5% had zero unplanned downtime. 10,918 websites saw no outage other than scheduled, pre-announced maintenance.
- 91.3% met 99.9% uptime or better. For context, 99.9% availability over a fully observed 30-day month corresponds to approximately 43.2 minutes of downtime..
- Infrastructure issues represented 37.8% of recorded unplanned incidents. 412 of 1,089 unplanned incidents (37.8%) were infrastructure issues. 565 (51.9%) came from website applications, configuration or customer actions.
- Most outages were caught by our monitoring first. In 741 of 1,089 unplanned incidents (68.0%), our alerts fired before any customer reported a problem.
About this data
The 12,480 websites in this report are a subset of the websites hosted on eWebGuru infrastructure. They are production websites that had active uptime monitoring with valid data for at least 20 days in September 2026. Websites without monitoring, or with too little monitoring data, are not included.
This is not a survey of the Indian hosting market. It shows how websites performed on one provider’s infrastructure, measured with that provider’s monitoring. Results will differ by hosting type, application, traffic pattern and configuration.
The websites run on eWebGuru servers in data centres in Mumbai and Delhi.
Uptime overview
Approximately two-thirds of included websites had no recorded downtime during their valid monitoring periods. Uptime figures in this section include planned maintenance, so a website is only counted at 100% if it had no downtime of any kind.
| Metric | September 2026 |
| Websites analysed | 12,480 |
| Median uptime (all downtime counted) | 100% |
| Websites with no downtime of any kind | 8,216 (65.8%) |
| Websites with planned maintenance only | 2,702 (21.7%) |
| Websites with any unplanned downtime | 1,562 (12.5%) |
| Websites at 99.9% uptime or higher | 11,400 (91.3%) |
Uptime distribution
| Uptime band | Equivalent downtime over a fully observed 30-day period | Websites | Share |
| 100% | None | 8,216 | 65.8% |
| 99.9% to below 100% | Up to about 43 minutes | 3,184 | 25.5% |
| 99.5% to below 99.9% | About 43 minutes to 3.6 hours | 764 | 6.1% |
| Below 99.5% | More than 3.6 hours | 316 | 2.5% |
| Total | 12,480 | 100.0% |
Note: Percentages are rounded independently and may not sum to exactly 100%.
Incident counts represent deduplicated operational events. Where one infrastructure event affected several websites, it was counted separately for each website. The same counting rule was applied throughout the analysis.
A total of 316 websites recorded availability below 99.5% during their valid monitoring periods. Over a fully observed 30-day period, 99.5% availability is equivalent to 3.6 hours of downtime.
What caused downtime
Infrastructure issues accounted for 412 of 1,846 incidents (22.3%). The largest share was planned maintenance, followed by issues at the website or customer level.
An incident here is one continuous downtime event. One incident on a server can affect several websites at once, which is why there are fewer incidents than affected websites.
| Cause | Group | Incidents | Share of all incidents | Median duration |
| Planned maintenance | Planned | 757 | 41.0% | 35 min |
| Infrastructure issue | Unplanned: hosting side | 412 | 22.3% | 28 min |
| Infrastructure issue or configuration issue | Unplanned: website side | 379 | 20.5% | 19 min |
| Customer-initiated action | Unplanned: application or configuration | 186 | 10.1% | 14 min |
| Other / unclassified | Unplanned: other | 112 | 6.1% | 41 min |
| All unplanned incidents | 1089 | 59.0% | 22 min | |
| All incidents | 1846 | 100.0% | Not reported |
Note: The overall median duration across planned and unplanned incidents is not reported. Category medians are shown separately and should not be averaged to calculate an overall median.
These percentages describe incident frequency, not the share of affected websites or total downtime. One infrastructure incident may affect multiple websites, while an application incident may affect only one.
Planned maintenance is reported separately because it is scheduled and announced in advance. It can still affect a business if it is poorly timed, so maintenance timing matters. [CONFIRM: state when maintenance windows run, e.g. late-night IST, and how much notice customers get.]
Within unplanned incidents alone, infrastructure issues were 37.8%. Website application, configuration and customer actions were 51.9%, and other causes 10.3%.
Detection and response
Our monitoring raised an alert before any customer report in 741 of 1,089 unplanned incidents. Early detection means our team can start work on an outage before the customer has to open a ticket.
In the remaining 348 incidents, the customer reported the problem first.
The median unplanned incident lasted 22 minutes. Website application and customer-initiated issues were the shortest to resolve (19 and 14 minutes). Infrastructure issues had a median of 28 minutes.
What businesses can do
Application or configuration issues and customer-initiated actions accounted for 51.9% of recorded unplanned incidents. The following practices address these risks, although incident frequency does not measure total customer impact. These steps reduce that risk, whoever hosts your website.
- Monitor more than the homepage. Watch the website URL and the services behind it: database, APIs, email and payment gateway. Assign a named incident owner and an escalation route so alerts are acknowledged and acted on.
- Keep tested, off-site backups. Automate them and test a full restore regularly. A backup you have never restored is not a recovery plan.
- Test changes in staging first. Application and configuration issues caused 20.5% of all incidents. Try WordPress, plugin, theme and WooCommerce updates on a staging copy before going live.
- Ask your host about maintenance. Find out how maintenance is announced, when it runs, whether it counts against the uptime commitment, and whether you can choose the window.
- Write down who does what during an outage. Decide who detects the issue, who tells customers, who restores service and who finds the root cause.
If you would rather not handle this yourself and for businesses considering managed hosting, review the monitoring, backup, maintenance and support coverage included in the selected eWebGuru plan.
Methodology
| Item | Definition |
| Reporting period | 1 to 30 September 2026 |
| Sample | 12,480 anonymised production websites. These are a subset of all websites hosted on eWebGuru infrastructure. |
| Included | Production websites with valid uptime-monitoring data for at least 20 days in the period |
| Excluded | Internal test sites, staging environments and websites with insufficient monitoring data |
| Uptime | Successful checks as a percentage of all valid checks, per website. Planned maintenance counts as downtime in the uptime figures. |
| Median uptime | The middle value when all 12,480 websites are ranked by their uptime percentage |
| Incident | One continuous downtime event. It may affect one website or several on the same server. Repeated failed checks during the same outage count once. |
| Planned maintenance | Downtime inside a maintenance window announced to customers in advance. It is included in uptime figures but excluded from unplanned-incident figures. |
| Customer-initiated action | Downtime caused by a change the customer made, such as a restart, migration or misconfigured update |
| Detected by monitoring first | Our alert fired before the customer contacted support about the incident |
Limitations. This report reflects eWebGuru’s monitored websites and monitoring method only. It does not represent all Indian hosting providers or all websites hosted in India.
About eWebGuru
eWebGuru has provided hosting in India since 2007 and serves more than 25,000 customers. It runs VPS, cloud and dedicated servers on its own network (AS133643) in CtrlS, Iron Mountain and Yotta data centres, and offers managed support for servers and public cloud.
Disclosure: This report was produced by eWebGuru using data from its own infrastructure and monitoring systems.
How to cite this report: eWebGuru, India Hosting Insights: Uptime Report, September 2026, published 29 Oct 2026.
Media contact: Ashok Arora, Founder & CEO, [email protected]
About The Author
