Since local nameservers exhibit overload, one may be tempted to run a private nameserver on each machine, and have it contact the global DNS hierarchy directly. This approach is more feasible as a backup mechanism than as a primary nameserver for several reasons. Using shared nameservers reduces maintenance issues, and the shared cache can be larger than individual caches. Not only does cache effectiveness increase due to capacity, but the compulsory misses will also be reduced from the sharing. With increased cache misses, the global DNS failure rate becomes more of an issue, so using private nameservers may reduce performance and reliability.
As a backup mechanism, this approach is possible, but has the drawbacks common to any infrequently-used system. If it is not being exercised regularly, failure is less likely to be noticed, and the system may be unavailable when it is needed most. It also consumes resources when not in use, so other tasks on the same machine will be impacted, if only slightly.