Skip to content
Discebo.dk

§ 19

Availability from MTBF and MTTR, with redundancy

How much downtime do we get, and what does an extra unit give?

Answer: The manufacturer specifies MTBF (mean time between failures) and you know yourself how long it takes to change or restart a unit (MTTR). From the two figures the calculator works out the unit's availability and what it becomes when two or three units stand parallel so the service runs as long as just one works. Downtime per year can be compared directly against the SLA.

Formula

A (én enhed) = MTBF ÷ (MTBF + MTTR)

A (n enheder parallelt) = 1 − (1 − A)ⁿ: tjenesten er kun nede, når alle n er nede på samme tid

Nedetid pr. år = (1 − A) × 8.760 timer

Symbols and units in the formula
SymbolMeaningUnit
mtbf_tMTBF (mean time between failures)timer
mttr_tMTTR (time to get it running again)timer
antalNumber of units in parallel—
=Availability with redundancy%
=Availability for one unit%
=Expected downtime per year with redundancys/år
=Expected downtime per year for one unittimer/år

Validity and source

Source
Tilgængelighed A = MTBF ÷ (MTBF + MTTR) · parallel redundans 1 − (1 − A)ⁿ (ITIL Availability Management, IEC 60050-192)

Does not apply when

  • The model assumes that units fail independently of each other and that the switch works. Common power supply, common software failures or wrong configuration hits both at once, and then redundancy doesn't help. The manufacturer's MTBF is a statistical average for many units, not a guarantee for yours.

Worked example

The calculator's own default values inserted. The figures are calculated with exactly the formula the tool uses.

Substituted

  • MTBF (mean time between failures)50,000 timer
  • MTTR (time to get it running again)8 timer
  • Number of units in parallel2

Giver

  • Availability with redundancy99.99999744 %
  • Availability for one unit99.9840 %
  • Expected downtime per year with redundancy0.81 s/år
  • Expected downtime per year for one unit1.40 timer/år

Members only

The calculator itself is locked

Create a free user account to unlock all articles, calculators, templates and range — and get access to quizzes and classes. It takes less than a minute: name, email and a password. No payment card.

Create free user →Log in

Want to try something first? See what you can use without an account — the junior zone, the encyclopedia, the videos and more.

Sådan bruger du den

Når en kunde vil have en tjeneste, der næsten aldrig er nede, skal du kunne vise hvad en ekstra switch, firewall eller strømforsyning gør. Beregneren regner tilgængeligheden fra producentens MTBF og jeres egen reparationstid og viser effekten af parallel redundans.

Step by step

  1. Find MTBF i enhedens datablad (timer).
  2. Skøn MTTR: hvor lang tid det tager at opdage fejlen, skaffe reservedel og få enheden i drift.
  3. Tilgængelighed for én enhed = MTBF ÷ (MTBF + MTTR).
  4. Med n enheder parallelt er tjenesten kun nede, når alle er nede: 1 − (1 − A)ⁿ.
  5. Regn nedetiden pr. år, og sammenlign med SLA'en.

Typical errors

  • At tro at to enheder fordobler sikkerheden. Fælles fejl som samme strømkreds, samme firmware eller en konfigurationsfejl rammer begge.
  • At bruge MTBF som levetid. 50.000 timer betyder ikke, at enheden holder i 5,7 år.
  • At glemme opdagelsestiden i MTTR. Uden overvågning kan en død enhed stå i dagevis.

Examples from everyday life

Kerneswitch i par

Kerneswitchen har en MTBF på 50.000 timer, og med reservedel på hylden tager et skift 8 timer. Du sætter to op i stack.

MTBF (mean time between failures)
50,000 hours
MTTR (time to get it running again)
8 hours
Number of units in parallel
2 · active/passive or active/active

It means: Én switch giver 99.984 % og en nedetid på 1.4 timer/år. To parallelt giver 99.99999744 % og kun 0.81 s/år, hvis de fejler uafhængigt.

Enkelt firewall uden reservedel

En filial har én firewall med MTBF 100.000 timer, men en ny skal bestilles hjem, så MTTR er 48 timer.

MTBF (mean time between failures)
100,000 hours
MTTR (time to get it running again)
48 hours
Number of units in parallel
1 · no redundancy

It means: Tilgængeligheden er 99.95202303 %, og nedetiden er ca. 4.2 timer/år. Med en reservedel på hylden falder MTTR, og nedetiden falder i samme forhold. En ekstra firewall i par gør endnu mere.

Dobbelte strømforsyninger i en server

En server har to strømforsyninger med MTBF 200.000 timer hver, og et skift tager 4 timer.

MTBF (mean time between failures)
200,000 hours
MTTR (time to get it running again)
4 hours
Number of units in parallel
2 · active/passive or active/active

It means: Med to forsyninger bliver tilgængeligheden 99.99999996 %, og den forventede nedetid er 0.01 s/år. Husk at sætte dem på hver sin strømkreds, ellers er de ikke uafhængige.

Rule of thumb

Halvér MTTR, så halveres nedetiden for én enhed. For to parallelle bliver den ca. en fjerdedel.

How to check the answer

Sammenlign med jeres egen driftslog: summér nedetiden over et år, og del med 8.760 timer.