Zero-Failure Option

West: Yes, two years ago, one of the archives was located in an offsite data center in downtown Nashville; there was a major failure of the power grid, which meant we lost connectivity to one of the hospitals, so that hospital centrally pointed all its imaging devices to the other node, and the radiologists were instructed to read from the second node. There is a bit of a manual process to it; there are some things like document scanners, for example, that won't automatically take a second destination. But in the case of CTs and MRs, you can put in multiple destinations, and we simply put out a notice saying the one PACS was down, redirect.

And the beauty is, when the second node comes back up, DICOM matching automatically takes place, so the replication is automatic. So in other words, it catches up with itself. It's almost like having side-by-side DVRs watching cable TV, and if you lose one of those boxes, you simply watch the other television.

HCI: What would your advice be for CIOs thinking about these storage and continuity issues?

West: The first thing would be for any CIO to determine if the scope of what they want to do with their imaging warrants this kind of configuration. This is not the sort of thing you would do with a 200-bed single community hospital.

HCI: It's somewhat surprising that more multi-hospital organizations haven't done this yet.

West: More and more have. But there's a geographic factor involved that we call the Goldilocks Syndrome, because everything has to be just right; if the two nodes are too far apart, as we discovered in one Ascension system, then creating the connectivity can be prohibitive.

HCI: So you have to be within the same community or region?

West: Yes, at least virtually. If you can afford the connectivity-and it's a minimum of a gigabit of connectivity-if you can put the two nodes within 100 miles apart, you could do this.

Healthcare Informatics 2011 April;28(4):32-36