TallyoCloud service level agreement (SLA)

Uptime commitment and how we measure it

Every plan, whether single-user, multi-user or 25+ users, carries the same monthly uptime commitment: 99.9% for each server, counted over each calendar month. A written agreement for 25+ users or a CA firm can add stricter terms, but never weaker ones. Plan details are on our pricing page.

• "Up" means you can work. Your server accepts a Remote Desktop login through our gateway and TallyPrime opens your companies.
• Downtime means nobody can. These are the minutes when no user can log in because of a fault in our servers, storage, network or gateway. Downtime starts when monitoring confirms the outage or when you report it and we confirm the fault is ours, whichever is earlier. It ends when logins work again.
• Measured from outside. An external monitoring service, running outside our own network, checks each server's login route around the clock. Its record, not our opinion, is what we use to calculate credits.
• Open to you. Ask for your server's monitoring report for any past month and we send it, with each confirmed outage and how it was counted.

A 30-day month has 43,200 minutes, a 31-day month 44,640 and February 40,320, or 41,760 in a leap year. The table shows how much downtime a 99.9% commitment leaves for each length of month. Anything beyond that allowance earns you a credit.

What does not count as downtime

An SLA is only as honest as its exclusions, so here they are in full. We keep them narrow: anything that goes wrong inside our servers, storage, network, gateway or data-centre power counts against us, including a single failed disk or switch.

• Planned maintenance that follows the rules below. If we break any rule, the time counts as downtime.
• Your side of the connection. Office internet, Wi-Fi, firewalls, VPNs, your device and the Remote Desktop app on it. Our internet speed guide helps you rule these out quickly.
• Services we don't run. The GST portal, the e-invoice and e-way bill systems, bank portals, and Tally Solutions' licensing and connected services.
• TallyPrime licence limits. If TallyPrime opens in Educational mode because more sessions are open than your licence has Tally Virtual Users, the server is still up.
• Changes you ask for or make. Software you install, settings changed with admin rights we gave at your request, or files your users delete.
• Suspension under our terms. For example, unpaid invoices after reminders, misuse, or a lawful order from an Indian authority, as set out in our terms of service.
• Force majeure. Events beyond reasonable control, such as floods, earthquakes, fire, war or a government-ordered internet shutdown. An ordinary hardware, power or network failure at our data centre is not force majeure.

Planned maintenance rules

Windows Server needs security updates, and we would rather patch on a quiet night than leave a known hole open. For planned work to be excluded from downtime, we must:

1. Announce it in advance on WhatsApp and by e-mail, with the date, start time, expected length and what will change.
2. Do it late at night, outside normal business hours.
3. Stay clear of the last days of every month, GST return due dates such as the 11th (GSTR-1) and the 20th (GSTR-3B), and the run-up to the tax-audit deadline. Our due-date calendar lists the dates we protect.
4. Finish on time. Any minute past the announced end counts as downtime.

Emergency security patches, for example when a flaw is being actively exploited, are announced as soon as they are scheduled. If we cannot warn you before we start, that time counts as downtime.

Service credits

If your server's uptime for a month falls below 99.9%, you get a service credit. The size of the credit depends on how badly we missed, measured as the downtime beyond that month's allowance.

• What the credit is based on. Our hosting fee for that server and month, meaning the per-user fees for everyone on that server, before GST.
• What it is not based on. Your TallyPrime licence or rental, TSS and extra TVU packs are paid to Tally Solutions or its partners, so they are outside our credits. One-off charges are outside them too.
• How it is paid. As a separate line on your next hosting invoice. Credits are not paid out in cash, and they never take away your right to cancel under our terms and refund policy.

An illustration of the rules on this page. The monitoring record for your server and our written reply decide the actual credit.

Support response targets by priority

Support 7 days a week. Requests are handled by priority, not in the order they arrive. Tell us how many people are affected and what they see; we set the priority from that, and you can always ask us to raise it.

We don't print a response time in minutes until we have measured it. Targets for each priority and plan will appear in this table once we have a full quarter of real tickets behind them, the same rule we follow for restore times. Step-by-step fixes for common problems are on our support page.

Backup and restore targets

Your server is backed up automatically: Daily automated backups, with 30-day retention, stored off the server in India. Our security page explains each backup layer, including the locked copy that ransomware cannot change.

• Recovery point. A restore brings your books back as they were at the backup point you choose. Entries made after that point need to be re-entered, so tell us the last voucher you remember.
• Restore process. We restore into a separate folder, never straight over live data. You check the trial balance and vouchers, then we swap it in or you copy across only what you need.
• Your own copy, any time. TallyPrime's own Backup and Auto Backup work on your server, and you can save a backup to your own computer without asking us. Our Tally data backup guide shows how.

Recovery point and recovery time targets in hours will be published here only once we have timed real restores in drills across a full quarter. A restore time nobody has measured is a guess, and we would rather show you the process than a number. Daily backups kept in India also help with the new daily-backup duty explained in our Rule 46(8) guide.

How we communicate during an incident

When something breaks, silence is the worst part. This is the sequence we follow for every incident that affects your server.

1. Detect Monitoring alerts us, or you do

   External monitoring flags the outage, or a customer reports it. Downtime is counted from the earlier of the two once we confirm the fault is ours.

2. Tell you A WhatsApp message to affected customers

   What is affected, what we know so far, whether any data is at risk and when you will hear from us next.

3. Update Updates at every step until it's fixed

   You don't have to chase us. Each change in the situation, including a revised estimate, reaches you on the same chat.

4. Resolve Logins checked and data verified

   We confirm users can log in and that companies open with their latest entries before we call it closed.

5. Explain A written incident summary

   The cause, start and end times, the downtime counted under this SLA, any credit due and what we are changing so it doesn't happen again.

Our legal reporting duties

As a cloud provider in India, we follow the CERT-In Directions of 28 April 2022: reportable cyber incidents go to CERT-In within six hours of our noticing them, and our system logs are kept in India for 180 days. If an incident touches personal data, our privacy policy sets out how we tell you and what we hand over so you can meet your own duties.

How to claim a service credit

1. After the month ends, WhatsApp us or e-mail tallyocloud@gmail.com with "SLA credit" in the subject, your server name, the month and roughly when you couldn't connect.
2. We check the external monitoring record and our incident log for your server.
3. We reply in writing with the measured uptime, the downtime counted, any exclusion we applied and the reason, and the credit due.
4. The credit appears as a separate line on your next hosting invoice.

If an incident summary we sent already shows the month crossing the allowance, it includes the credit calculation, so you don't have to ask. Disagree with our figures? Ask for the raw monitoring record. If we still can't settle it, our Grievance Officer handles the complaint; the details are on our contact page.

How to read any Tally cloud SLA

Whether you choose us or another host, these six questions separate a real Tally cloud SLA from a percentage on a banner.

• Per server or overall?A network-wide average can hide a week-long outage on your server. Ask for a per-server figure.
• Who measures it?Monitoring from outside the provider's network is harder to argue with than its own dashboard.
• Rules for maintenance?Look for advance notice, night-time slots and protected due dates, not "excluded at our discretion".
• Credits on what?Check whether the credit is a share of the fee you actually pay, and how you claim it.
• Support priorities in writing?A written order of priority beats an unmeasured "24×7" promise.
• Backups and exit?Retention, restore process and your right to a full copy when you leave.

For reference, Tally Solutions' own TallyPrime Cloud Access page and FAQ did not state an uptime commitment or data-centre region when we checked on 6 October 2026. Our side-by-side is in TallyPrime Cloud Access vs Tally on Cloud, and the full buyer's checklist is in how to choose a Tally on Cloud provider.

Changes to this SLA

This SLA forms part of our terms of service. We may improve it at any time. A change that reduces your commitments never applies in the middle of a term you have paid for: it takes effect from your next renewal, and we e-mail you before it does. The date at the top of this page shows the latest version, and earlier versions are available on request.

Frequently asked questions