i couldn't find any specific requirements for distance between main site and DR at the regulations or best practices only descriptions as “safe distance” can any one provide reference to something some specific ?

  • 1
    It depends what you're protecting against. Fire: a mile is probably fine. Country falling into civil war: you need somewhere in a different country. – Matthew Feb 9 '17 at 7:06

It depends on the geographic scope of the threat that you are trying to mitigate balanced with the costs and impact of distance on operations and recovery.

If the main site is in a drought-prone wooded forest with dry grasslands, you'd want your DR site beyond the fire zone. If the main site is in an earthquake zone, you'd want your DR outside of that zone. If the main site is in an area where the infrastructure itself is a risk (unstable governments, unreliable communication lines, etc.) you would want the DR site in an area with different infrastructure. And so on.

But distance incurs cost and impact. To place your DR site on the other side of the planet means that you are maximally separated from any geographic zones the main site might be exposed to, but in a disaster, it is maximally difficult to get to the DR site. Distance, itself, becomes a risk factor to consider and weigh against the risks calculated for the main site.

So, I doubt you would find any regulations regarding a "safe distance" because it is a function of the geographic risks, risks incurred by increased distance, and costs and availability of a DR site in a particular location.

Always remember, mitigations always introduce their own, new risks to weigh. Costs and risks of mitigations must not exceed the impacts of the main risk subject.

| improve this answer | |

In addition to the previous answer, there is another consideration.

If you want your DC's to use data synchronisation between databases for high availability on the DB's, you need to make sure that they are close enough together that the latency is within the tolerances set by the vendor.

If I remember correctly, for Oracle, that used to be around 20 miles or so up to around 40 - though it has been a while since I thought about it so I might be off there. In addition, it is possible that newer networking capabilities may have changed the requirement.

Either way, latency between the DC's is important if you are wanting true, real-time data synchronisation that ensures no data loss at all if one of the DC's or one of the key services in DC goes down. This is also important if you are load-balancing and need real-time or near real-time data sync.

At the opposite extreme, if you do want DC's in a different country to protect from geographic issues, you also need to take into account differing regional laws regarding data protection, etc.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.