For a hospital with roughly 1,000 users and a big push toward cloud services, would you still run DHCP and DNS locally? What would drive that decision?
Would you keep DHCP and DNS on site for a hospital moving to the cloud?
Answers
Yes, I'd keep both local. A redundant internet setup helps, but a major incident can take out connectivity while the hospital is dealing with the same incident. DHCP lets devices obtain or renew addresses, and local DNS can keep resolving the names needed by services that remain on site.
That doesn't make cloud applications available during a WAN outage. Map out which clinical and operational functions need to keep working, including their authentication and application dependencies, then test the whole chain with connectivity down.
Reply to this answer
Start with the business continuity and disaster recovery plans, not a blanket rule about cloud versus on premises. For each service, work out what breaks if DHCP, DNS, authentication, the WAN, or a cloud application is unavailable.
The size of the site is a useful warning sign, but the actual decision depends on the required recovery time, the services that can operate locally, and what downtime the hospital can accept. Test those assumptions rather than just checking that a second circuit exists.
Reply to this answer
Local DHCP is the straightforward call. DNS needs a closer look at what clients actually need to resolve. A local cache may help with previously resolved names, while replicated local zones can provide answers for the internal services they host.
For a site this critical, I'd consider at least two local DNS servers and offsite replicas, with zone replication and failure behavior tested. Also check whether local authentication or network access control is needed for the workflows expected to survive a WAN outage.
m.257!torvexa:
So local DNS alone wouldn't help if the application and authentication it points to are both cloud-only?
7w~cloverpiston_316:
Right. Resolving the name is only one dependency. The service and whatever it needs for access also have to be reachable.
Reply to this answer
DHCP could run on suitable network equipment rather than a separate server, depending on the resilience and management features you need. For DNS, I wouldn't automatically put it on site if the risk assessment shows nothing useful can run without the WAN anyway. That's a different call from saying DNS isn't critical.
_v39+quendixa.2:
That distinction matters. Check the local equipment and downtime workflows before deciding there is nothing useful left to resolve.
Reply to this answer
Beyond core network services, consider a local update or content cache for a site this size. It can reduce repeated downloads over the WAN, though it's a separate capacity and operations decision from keeping clinical workflows running during an outage.
Reply to this answer
Add your answer
Answers are reviewed before publication.
_b69+fendari.8:
Include clinical staff in that exercise. They can help identify which workflows need a workable fallback, not just which servers appear on the network diagram.