When I hear "Mission Critical", the first thing that instantly comes to mind is "OpenVMS"!

The meaning of 9s (eg. 99.999 %) does not automatically imply mission critical. This are two different issues. They  often go together since high availibility and reliability supoorts mission critical applications more smoothly.

One is true in any case: openVMS is first in line and a couple of reinventors could gladly avoid the mistakes of the early days. VMS collected so much experience on the topic over 30 Years. So, its worth cribing concepts still and blare out in bright (new) colors.

Keep in mind: mission critical for a Sherpa might allow 0% availability of any kind of IT (even nower days).

OpenVMS clusters comes to mind for mission critical applications, using cluster partners in different locations. For systems not in need of that much robustness, local clusters. Mission critical systems should be available, secure and responsive 100% of the time they are needed (though they may not necessarily be needed for 24x7 service).

cool blog

I bookmarked this link. Thank you for good job!,

Great. Now i can say thank you!,

The most forgotten aspect of the five 9's is the quality of (application) code running the five 9's applications.

Having all this "infrastructure" availability is not much use if the code contains flaws, behaves unpredictably or whatever. See the fact that the likes of Google, Twitter, Facebook etc. have been off the network for minutes and hours, but nobody was seriously affected. Well, those that were using Google applications were, I suppose.

Whilst the platform is of relevance, as is the infrastructure, it is of no use if the (application) code breaks.

So, the question is: when is a bug a "nuisance" and when is it not?

One definition might be that when human lives are at stake, then bugs are not simply a nuisance.

What has any of the above go to do with five 9's? If it is not obvious by now, then...we do not need them.

It seems - once again - that BCS-marketing still has to do some homework on the history of  

Where you require system availability measured in multiple years, nothing beats OpenVMS clusters: there are several that measure uptime over 10 years, not a single second that the system wasn't available for the users, even when moving hardware around, or updating hardware and software; BUILT IN for over 30 years.

The same may be said of NonStop but I have no fugures on that. But the state of that OS is impressive as well.

Both systems are now owned by HP - but marketing does a bad job (for both HP and their customers) forgetting these systems.

"Available"  means "Accessable when required". In most cases, this does not need to be 24*365.25 (or as close to that number as possible): For a teenager that wants to chat with friends, the laptop with internet access should be at hand, but may be switched off otherwise. But it's very different where Bad Things Happen if any component fails. be it hardware (network, computer, storage) or software (OS, application) or data.

From a user perspective, In either case the number of nines should be as high as possible - for the time of the requirement.