A 99.95% uptime SLA permits roughly 22 minutes of unavailability in a 30-day month. A 99.9% SLA permits roughly 43 minutes. The gap between them is about 21 minutes a month, and the number is the least interesting part of the contract. What separates two providers is not the percentage they print but what sits underneath it.
Availability percentages are calculated, not observed. Someone decides what counts as unavailable, which components count, and which periods are excluded before the arithmetic starts. Two providers can both be honest and still produce different numbers for the same month. Reading the definition takes ten minutes, and it tells you more about what you are buying than any comparison of headline figures ever will.
Read the exclusions first
Almost every provider excludes scheduled maintenance. If maintenance is not excluded, the window matters more than the percentage. If it is excluded, ask how much of the month is reserved for it. A provider with a generous maintenance allowance and a strict measurement method can deliver less usable uptime than one with a tighter window, even when both print the same number on the first page.
Then look at what else is carved out. Force majeure, upstream carrier failures, customer-caused outages and faults in components the provider does not operate are commonly excluded. Some of those exclusions are reasonable. The question is whether, after all of them are applied, anything meaningful remains inside the promise you are being asked to rely on.
- Is scheduled maintenance excluded, and how many hours are reserved for it
- Are upstream carrier faults excluded, and who decides that a fault was upstream
- Are single-component failures excluded when the service is sold as a whole
- Is measurement continuous, or does it stop during declared events
Then read the remedy
An SLA is not a service level. It is a remedy attached to a service level. The remedy tells you what happens after the failure: a service credit, a refund, or an apology. Read it before you compare anything, because the remedy is where providers differ most, and it is the only part of the contract you will actually use.
- What is the service credit, and is it a credit or a refund
- Does the credit require you to file a claim, and within what window
- Does the SLA cover the whole service or only a component
- Is the measurement per month, or over a rolling period
- Is the credit capped, and at what fraction of the monthly fee
The claim requirement deserves particular attention. A credit that is applied automatically is worth more than a larger credit you must remember to request inside a thirty-day window while your team is still recovering from the incident. Ask who files the claim, how they learn that an incident occurred, and what evidence the provider requires. If the answer is that the customer must notice and apply, treat the credit as a discount you may never receive.
Also ask what the credit is calculated against. A credit measured against the affected component, rather than the monthly bill, can be small enough to be invisible. A credit capped at a fraction of one month is a refund of a fraction of one month, not compensation for the outage. Neither arrangement is dishonest, but only one of them is likely to be described accurately in a sales conversation.
How we measure and report
Our availability commitment is 99.95%, measured continuously across the service rather than per component, and reported monthly. Planned maintenance is scheduled in MYT business hours and notified at least seven days in advance, so it is visible before it happens rather than discovered afterwards. When something does go wrong, we acknowledge support requests within fifteen minutes and publish a postmortem within five business days.
The postmortem is the part customers tell us they value most, and it is the part that is not in the contract. It sets out what failed, how it was detected, how long recovery took, and what we have changed so that the same failure does not recur. An SLA tells you what you are owed after an outage; a postmortem tells you whether the outage will happen again. That is not a contractual obligation, and we think it is the more useful of the two.
We run the infrastructure ourselves, in Malaysia, with redundancy designed so that a single failure does not become an outage. Dual power feeds, generator backup, cooling sized for one unit failure and diverse network paths are the boring parts of the design, and they are the reason an availability figure is achievable rather than aspirational. Monitoring runs continuously and is tied to the same measurement we report to customers.
Then compare the number
Two providers quoting 99.95% can be offering very different things. One may exclude maintenance and cap the credit; the other may measure continuously and refund automatically. The percentage is the same; the contract is not. Comparing the numbers before reading the definitions is the most common way buyers end up disappointed by a service that is performing exactly as documented.
- Same percentage, different exclusions
- Same percentage, different remedy
- Same percentage, different measurement period
An SLA is a promise about what happens after the failure, not a prediction that it will not.
What an SLA does not cover
An SLA does not cover the cost of the outage itself. It does not cover the staff who spend a night restoring a service, the orders that were not taken, or the customer who decides not to come back. Those costs are usually several orders of magnitude larger than the credit, which is why the credit should never be the reason you choose a provider.
What you are really buying is the engineering behind the number: the redundancy, the monitoring, the maintenance discipline and the honesty of the reporting. The percentage is a summary of those things, not a substitute for them. Ask for the postmortems, ask how maintenance is scheduled, and ask who is on call at three in the morning. The answers will tell you more about your 22 minutes than any contract clause.
If the answers are satisfactory, the rest is straightforward: billing in Ringgit, payment through FPX or TnG eWallet, data held in Malaysia under Malaysian law including the Personal Data Protection Act 2010, and support during MYT business hours. Those are the commitments that sit underneath the percentage, and they are the ones worth reading closely.
