IT Glossary
What is Uptime?
The percentage of time a website or service actually works: the number — with its arithmetic of "nines" — measuring reliability promised and delivered.
Uptime is reliability expressed as a single figure: the share of time a service — website, store, application — is actually working; its complement, downtime, is the time it is not (during which the store sells nothing, the form receives nothing, and the company does not exist online). Feel the arithmetic of the nines once and inflated promises stop impressing: 99% monthly permits over seven hours of outage per month (a lot); 99.9% about 43 minutes; 99.95% about 22; 99.99% under five — and every additional nine costs exponentially more infrastructure (redundancy, failover, no single points of failure), which is why reliability is purchased against real need: what an hour of downtime costs you, times the probable hours, equals the rational reliability budget. The administrator's practice comes in three pieces. Independent monitoring: never rely on the provider's self-reported figure — external tools (free tiers upward) probe the site from the internet every minute and alert you the moment it drops; they cost next to nothing and give you both your own number and the winning argument in any SLA dispute ("according to our independent measurements..." changes the tone of the conversation). Reading guarantees properly: contractual uptime comes with definitions (what is measured, what is excluded — announced planned maintenance is a normal exclusion, elastically defined "external causes" are a hole; and over which window, since a monthly versus yearly guarantee drastically changes what a four-hour outage means) and with remedies (service credits compensate symbolically, never the real business loss — the reason choosing a provider on verifiable history beats negotiating penalties). And internal hygiene: downtime does not come only from hosting — an expired certificate, an unpaid domain, a failed update or a dead integration produces a "down site" on perfectly healthy hosting, so complete monitoring also covers content checks, certificates and critical flows. The operational maturity question stays simple: last time the site went down, did you learn it from your systems or from a customer?
Let’s talk about your project
Message us on WhatsApp or send an email — you talk directly to a developer.
office@northdan.com · +40 752 070 247
Why it matters for your business
Outages learned from systems, in minutes
External monitoring with alerts turns "since when has it been down?" into an intervention starting at minute two — lost sales counted in minutes, not hours.
Your own number, not just the provider's
Independent measurement builds your own uptime history — the objective basis of any SLA discussion and of any decision to switch providers.
Reliability bought on arithmetic
Knowing what an hour of downtime costs and what each nine costs, you size guarantees by calculation — neither underinsured on the store that sells nor paying hospital-grade redundancy for a brochure site.
Frequently asked questions
What uptime is "normal" to expect from decent hosting?
Realistically, by tier: good shared hosting delivers about 99.9% in practice (written guarantees vary); VPS and managed cloud 99.9-99.95%; seriously redundant infrastructures 99.99%+ with architecture to match. Sustained performance below 99.9% — more than about 43 minutes down, month after month — on a decently paid service is a signal to move. The nuance that matters: your own measured figure over time beats any percentage on a sales page, so start monitoring before you need arguments.
The site "works for us" but customers say it will not load — how do I reconcile the versions?
The classic distributed diagnosis: a partial outage — by region, by carrier, by DNS, or only on certain flows (the site opens but payment fails) — escapes a check from your own office. Hence the value of monitoring from multiple locations (serious tools probe from several countries and networks) and of flow checks (not just "does the page respond" but "can an order be completed"). Until you have tools: one quick test from a mobile connection instead of the office WiFi and one through an online "is it down" service settle in two minutes whether the problem is theirs, yours or somewhere on the way.
Does planned maintenance count as downtime?
Contractually, usually not — if announced per the SLA (window, notice period) and done in the agreed intervals; commercially, your customers do not read your provider's SLA: a store "under maintenance" on Saturday at noon loses sales whatever the classification. Healthy practices: maintenance windows in your business's genuinely dead hours (not the provider's), architectures allowing updates without stopping (standard on modern platforms — ask before choosing) and an honest maintenance page with an estimated return time for the few affected.
Let’s talk about your project
Message us on WhatsApp or send an email — you talk directly to a developer.
office@northdan.com · +40 752 070 247