I once hosted a Zope site on three different datacenters - one for the ZEO server (the object store) and two others for the clients (the front-end/app logic). Worked well enough (Zope caches things very efficiently).
No. I really didn't needed to do it - I just wanted to know if it would work. I suspect a similar approach with light front-ends on low latency regions with heavy loads on cheaper regions would be a good technique depending on your problem.
Brazil's once largest airline used Zope a lot and not even they had problems that needed solving this aggressively.
Now that I am writing this, I remember there was a government (Plone-based) site that suffered a major data-center outage and we quickly switched to running their Varnishes in Brasilia off our mirror in São Paulo for weeks. Nobody outside the technical team noticed.
No. I really didn't needed to do it - I just wanted to know if it would work. I suspect a similar approach with light front-ends on low latency regions with heavy loads on cheaper regions would be a good technique depending on your problem.
Brazil's once largest airline used Zope a lot and not even they had problems that needed solving this aggressively.
Now that I am writing this, I remember there was a government (Plone-based) site that suffered a major data-center outage and we quickly switched to running their Varnishes in Brasilia off our mirror in São Paulo for weeks. Nobody outside the technical team noticed.