CERT-In Compliance: The 6-Hour Rule and What to Build First

The clock starts when you notice, not when you're ready
At 11:40 PM on a Tuesday in March 2025, a 60-person logistics firm in Peenya noticed their file server had been encrypted. The IT lead did what most people do: he called his MSP, isolated the machine, and started restoring from a Veeam backup taken the night before. By 3 AM, systems were back. He went to sleep thinking the worst was over.
Fourteen months later, the company received a show-cause notice from CERT-In. They had failed to report a ransomware incident within six hours. The notice asked for logs, an incident timeline, and an explanation for the delay. The company had no formal incident register, no NTP-synced timeline, and had overwritten the affected VM during restoration. Penalties under the IT Act's Section 70B framework can run into lakhs, but the real cost was the eight weeks of management time spent reconstructing events from WhatsApp messages, plus a client audit that froze a ₹1.2 crore contract for two months.
The rule is simple to state and hard to build for: any cyber incident must be reported to CERT-In within six hours of noticing it. Most Indian SMEs cannot meet that today. Not because they don't care, but because nobody has sat down and engineered the plumbing — time sync, log retention, escalation, evidence preservation — that turns a six-hour deadline from a panic into a checklist.
This article covers what the 2022 directions (and subsequent FAQs) actually require, what to build before an incident, and where the common failure points are.
What CERT-In actually requires (and what it doesn't)
The directions were issued on 28 April 2022 under Section 70B(6) of the IT Act, and they came into force on 27 June 2022. They apply to every organisation using a computer resource in India — service providers, intermediaries, data centres, body corporates, and government organisations. In practice that means every company with an office network, a cloud subscription, or a payroll system.
The core obligations are:
- Report cyber incidents within six hours of noticing them, via email to incident@cert-in.org.in or the CERT-In portal, in the prescribed format.
- Maintain logs for 180 days rolling, within Indian jurisdiction.
- Sync ICT system clocks to NTP servers maintained by NIC or NPL.
- Maintain KYC and subscriber records for specified services (VPNs, cloud, virtual asset service providers).
- Provide CERT-In with information on demand, including logs and system details.
- Designate a point of contact for CERT-In coordination.
What is not required, despite what some vendors will tell you: there is no mandated SIEM product, no specific log format, no requirement to buy a particular appliance. The directions are outcome-based. You can meet them with open-source tools, a mid-range FortiGate, a Synology NAS, and disciplined process — or you can spend ₹40 lakh on a branded stack and still fail the six-hour test because nobody documented who is supposed to send the email at 2 AM.
The six-hour rule, precisely
The six hours start when the incident is noticed, not when it is confirmed. This is the part most companies get wrong. If an employee reports a suspicious popup at 10:00 AM and the IT team spends the day investigating before calling it an incident at 4:00 PM, the clock started at 10:00 AM. The reporting window closed at 4:00 PM.
CERT-In's FAQs acknowledge that at the six-hour mark you may not have full details. That is fine. You report what you know: affected systems, apparent nature (ransomware, unauthorised access, data breach, phishing with compromise), initial indicators, and a point of contact. Follow-up details go in later updates. The mistake is waiting for a complete picture.
A useful discipline: separate "reportable incidents" from "events." A phishing email that was blocked by Microsoft Defender for Office 365 Plan 2 and never resulted in a click is an event. A phishing email where one user entered credentials and the account was accessed from a new IP is a reportable incident. Write this distinction down and train your L1 team on it. Without it, either everything gets reported (and your team stops bothering) or nothing does (and you get a notice 14 months later).
180-day log retention in Indian jurisdiction
Logs must be retained for a rolling 180 days and must reside in India. This catches out companies who use global SaaS logging tools with US-East storage. If your only copy of authentication logs sits in Splunk Cloud's US region, you are non-compliant even if you can pull the logs on demand.
What "logs" means is not exhaustively defined, but the practical minimum for a 20-500 person company is:
- Firewall traffic and session logs (source IP, destination IP, port, action)
- VPN authentication and session records
- Domain controller security event logs (4624, 4625, 4648, 4672, 4768)
- Endpoint detection and response (EDR) telemetry
- Email gateway logs (message tracking, delivery, spam verdicts)
- Cloud control-plane audit logs (Entra ID sign-ins, AWS CloudTrail, Google Workspace audit)
- Database audit logs for systems holding personal data
Retaining all of these for 180 days is not expensive if you size it properly. A 100-user office generating firewall and endpoint logs at a typical rate stores roughly 400-800 GB over 180 days when compressed. A Synology DS923+ with four 8 TB drives in RAID 5 (₹1.1-1.3 lakh for the unit and drives, 2026 pricing) covers it comfortably. What adds cost is the ingestion layer if you go with a commercial SIEM.
| Log source | Typical daily volume (100 users) | 180-day compressed | Where to keep it |
|---|---|---|---|
| FortiGate firewall (FortiGate 90G) | 1.5-3 GB | 120-250 GB | On-box or syslog to NAS |
| Windows Domain Controller Security log | 0.5-1 GB | 40-90 GB | WEC to NAS |
| Microsoft Defender for Business telemetry | 0.3-0.8 GB | 25-70 GB | Defender portal (data residency: India region) |
| Sophos Central / EDR | 0.3-0.7 GB | 25-60 GB | Vendor cloud, India region |
| Microsoft 365 Unified Audit Log | 0.2-0.5 GB | 18-45 GB | Purview (set retention policy to 180d) |
| File server / NAS access logs | 0.3-0.6 GB | 25-55 GB | Synology Log Center to NAS |
| Total | 3-6.6 GB | 250-570 GB | — |
Microsoft 365's default Unified Audit Log retention is 180 days for E3 and E5, and 90 days for Business Standard. If you're on Business Standard and storing nothing else, you are one subscription tier away from a gap. Enable a 180-day retention policy in Purview explicitly and verify it quarterly.
NTP sync is not optional
Every ICT system clock must be synchronised to NIC or NPL NTP servers. This is a one-line configuration and yet it is the single most common source of confusion during an incident review. If your firewall logs show the breach at 02:14:33, the domain controller says 02:14:11, and the endpoint says 02:13:58, your six-hour timeline is meaningless — to CERT-In and to your own forensic review.
Configure Windows domain controllers to sync from nic.ntp.gov.in or time.nplindia.org, then propagate via GPO to every joined machine. Network devices (FortiGate, Cisco, Sophos) should sync from the same sources. Servers on Linux — chronyd pointing at the same upstream. Cloud VMs: AWS Time Sync Service or Google's metadata.google.internal are fine for internal sync, but the on-prem appliance that reports to CERT-In should anchor to the Indian NTP pool so your reporting timestamps are defensible.
Test it. Once a quarter, run w32tm /stripchart /computer:nic.ntp.gov.in /samples:3 on the PDC and check the offset is under 100 ms. If it isn't, port 123 outbound is probably blocked on your firewall — a very common mistake when the NGFW rules were written years ago and NTP was never added.
KYC records: who this actually applies to
KYC is where most SMEs can relax. The requirement to maintain KYC and subscriber records applies mainly to data centres, virtual asset service providers, VPN service providers, and cloud service providers. A 60-person manufacturer running its own server room and Office 365 is not a "service provider" in the relevant sense for KYC.
But there are two catches. First, if you have a captive VPN for remote staff, you need to keep five-year records of subscriber names, physical addresses, and validity periods. Many companies run a FortiClient VPN or a Sophos SSL VPN and never recorded this beyond an HR file. With DPDP Act enforcement now active in 2026, this overlaps with consent and retention requirements — build a single register.
Second, if you offer any service to external customers over the internet — even a customer portal — and you store their data, you may fall under the "intermediary" or "body corporate" definition and need to designate a point of contact. The threshold is not users; it's the nature of the service.
Trade-offs: what this costs and where it doesn't fit
Honest numbers. For a 50-user Bangalore office building CERT-In readiness from scratch in 2026, you're looking at:
| Item | Capex | Opex (annual) | Notes |
|---|---|---|---|
| FortiGate 90G (or similar NGFW with logging) | ₹1.6-2.2 lakh | ₹35-55k (FortiCare) | 90G has onboard storage for logs |
| Synology DS923+ + 4×8 TB drives (RAID 5) | ₹1.1-1.3 lakh | ₹15-25k (disk replacement, power) | Log retention target |
| Windows Event Collector server (or existing server role) | ₹0-1.5 lakh | — | Often reuses existing hardware |
| EDR: Sophos Intercept X or Microsoft Defender for Business | — | ₹900-2,200 per endpoint/year | Defender included in Business Premium |
| Microsoft 365 Business Premium (audit log retention, Purview) | — | ₹1,850-2,400/user/year | Necessary for 180-day audit log |
| NTP (NIC/NPL, on-prem anchor) | ₹0 | ₹0 | Free if you configure it |
| Process build: incident register, escalation matrix, reporting template | ₹40-80k one-time | ₹0 | Real work, not a product |
| One-time + Year 1 total | ₹3-6 lakh | ₹2.2-3.5 lakh | — |
Where this does not fit: if your entire infrastructure is a single Office 365 tenant with five users and no servers, you do not need a FortiGate or a NAS. You need M365 Business Premium with 180-day audit retention, NTP-synced laptops, a documented incident register, and a one-page reporting workflow. Total cost: your existing Microsoft subscription plus a good afternoon's work. Buying a ₹2 lakh firewall for a five-user company is the wrong instinct.
The second honest caveat is about SynergyScape itself. If you already have a capable internal IT team and a working SIEM, you don't need our managed service to build this — you need the checklist below and someone to force the discipline. We charge for the discipline, not for a secret product. If you have fewer than 20 endpoints and no regulatory client audits, a basic M365 configuration plus a template may be sufficient. If you're listed on a client's supplier register and asked for a CERT-In compliance statement, that's when the build pays back.
Building the reporting workflow before you need it
The six-hour clock demands three things: detection, classification, and communication. All three can be ready in advance, and none of them require a vendor.
Step 1: Designate an incident commander and a backup
One person, by name, is accountable for determining whether an event becomes an incident and, if so, sending the CERT-In email. Their backup is named too. Their phone number is on the on-call rota. Their mailbox has the reporting template bookmarked. If your incident commander is the CTO, you have a problem on holiday weekends.
Step 2: Build a two-page reporting template
The email CERT-In wants is not a forensic report. It is: who you are, what happened, apparent scope, initial indicators, containment so far, and contact details. Draft this as a Word template with placeholders. Have it reviewed by your legal team once. Store it in a shared folder your incident commander can reach from a phone.
Step 3: Define what counts as reportable
For a typical Indian SME, these are the reportable categories:
- Ransomware on any production system
- Unauthorised access to a server or cloud account (confirmed)
- Data breach of personal or customer data with potential exposure
- Compromise of an executive or finance mailbox
- DDoS attack affecting external services
- Phishing campaign that resulted in credential use
And these are usually not:
- Phishing emails that were blocked or reported without a click
- Failed login attempts
- Spam or unwanted email
- Vulnerability scan traffic
Put this list on a single page. Train L1 and L2 twice a year. When the clock starts at 10:00 PM on a Saturday, you need a junior engineer to be confident enough to escalate without waiting for a senior.
Step 4: Preserve evidence before you remediate
The Peenya logistics example above failed because the VM was destroyed during restoration. Before you wipe, snapshot. Before you rebuild, export the disk. Veeam Backup & Replication v12 can take a VM snapshot in 30 seconds; a forensic disk copy of a 500 GB VM to a dedicated NAS volume takes 15-25 minutes. That is a cheap insurance policy. Do it before you restore, every time. Put the instruction in the incident runbook.
Step 5: Test the clock
Once a quarter, run a tabletop. Pose a scenario ("file server encrypted, 25 users affected, EDR alert at 08:15 Tuesday"). Give the incident commander 30 minutes to fill the CERT-In template with placeholder data and email it to your internal audit address. The first one will take 45 minutes. The fourth will take 12. That improvement is what you are buying.
The log retention problem nobody wants to talk about
Retention is cheap to provision and expensive to operate. The failure mode is rarely disk space. It is that the logging pipeline quietly stops, and nobody notices until an incident six months later when you reach for 180 days of data and find 47.
Common causes:
- Firewall syslog stops forwarding after a firmware upgrade because the destination port reverted to default
- Windows Event Collector subscription expires at 180 days because nobody extended it
- NAS fills up because the retention policy was set incorrectly (append-only, no rotation) — this can happen in six weeks, not six months
- Microsoft Purview retention policy set to 180 days but the underlying audit log was paused because someone with the right role toggled it
Build a weekly check: count logs received per day for the last seven days, compare to the previous seven. If the gap exceeds 30%, something broke. Automate an alert to the IT manager's email. This is a 20-line PowerShell script; write it once and schedule it.
What to do this month
You cannot build all of this in a week. You can order the work in a way that front-loads the six-hour risk.
Week 1 — NTP. Sync everything to NIC/NPL. Test the offset. Fix the firewall rules. This is the foundation for everything else and costs nothing but an afternoon.
Week 2 — Point of contact and template. Designate an incident commander and backup by name. Draft the CERT-In email template. Publish a one-page escalation matrix.
Week 3 — Retention audit. Look at every log source in the table above. Ask: where is it stored today, where will it be stored in 180 days, and is that in India? Fix the gaps. Order the Synology if you need one — lead times in Bangalore are typically 2-4 days from authorised distributors, so no excuses.
Week 4 — Tabletop. Run the first one. Record time-to-report. Identify what stalled. Repeat quarterly.
This is where our managed services team usually gets called — not to buy a product, but to run the audit and write the runbook while your staff focuses on their day jobs. If you want to walk through your specific setup, get in touch and we'll map it against the CERT-In checklist in a 90-minute session.
FAQ
Does CERT-In apply to small companies?
Yes. The directions apply to every organisation using a computer resource in India, without a size threshold. A 10-person company running Office 365 and a firewall is in scope. Enforcement is more common where there are client audits or regulatory exposure, but the obligation is universal.
How many hours do I have to report an incident?
Six hours from the moment the incident is noticed, not from when it is confirmed. If an employee reports a suspicious event at 10:00 AM, the clock starts then. You can report with partial information and update later.
Do logs have to be stored in India?
Yes. The directions require logs to be maintained within Indian jurisdiction for 180 days. Many global SaaS logging tools default to US or EU regions. Check your retention location — moving to an India region is usually a configuration change, not a re-platforming.
What is the penalty for non-compliance?
The IT Act's Section 70B framework provides for penalties determined by the government, and CERT-In can issue notices, require remediation, and escalate to other regulators. Actual fines vary; the bigger cost is usually contractual — client audits, supplier-register suspensions, and management time.
Does Microsoft 365 meet CERT-In log retention automatically?
No. Business Standard retains audit logs for 90 days. Business Premium and E3 default to 180. You must explicitly configure a retention policy in Purview for 180 days, verify it, and monitor it. Note that Business Premium's audit log is stored in-region for Indian tenants, but the setting is not on by default at the retention period you need.
What NTP servers should I use?
nic.ntp.gov.in or time.nplindia.org. Configure domain controllers to sync from one of these, propate to joined machines via GPO, and point network devices directly. Test with w32tm /stripchart quarterly. Do not rely on time.windows.com — it works for general sync but is not the reference the directions specify.
Next action
Open your laptop, run w32tm /stripchart /computer:nic.ntp.gov.in /samples:3 on your primary domain controller, and note the offset. Then open your firewall's syslog settings and check where logs are going. Those two checks take five minutes and tell you whether your six-hour clock is defensible or fictional. If the answer is the latter, the managed services team can help you build the workflow before you need it.
