I think dual- or multi-site operation without backup power aims to provide the same service continuity by using load balancing and excess processing, storage and networking capacity while removing the UPS and generator. With rapid workload mobility provided by virtualization and good load balancing, this should work fine in certain situations. There may be some performance issues, though, introduced by distances between sites. I can see this working better for multiple sites, where a site's worth of excess capacity is distributed among several locations and is used to accommodate the extra workload in case a site fails. With dual sites, each site would have to have double its normal capacity. As with power and cooling infrastructure, N+1 is more affordable for larger values of N.
The real test may be simple economics: Is the cost of UPS, batteries, generator, electrical wiring and switching infrastructure and maintenance required for backup power greater or less than the cost of extra servers, disks, networking gear, cabling, software licenses and maintenance required for the excess capacity?
I'd love to hear from anyone who is using this strategy successfully.
Ken
__________________
DR/BC Planner
|