Both the technical and administrative infrastructure of the institutional electronic mail service should exhibit the property of reliability. Reliability is concerned with the accessibility, predictability, and error rate of the overall service. Reliability is a key characteristic of a large-scale, production institutional electronic mail service in an increasingly novice client environment, where clients expect ``black telephone'' service.
In order to provide a system that incorporates reliability
as a property of the architecture,
rather than as pleasant operational accident,
the service critical components
must be provisioned
by a single, central operational authority.
By central operations,
we mean that
the components of the technical infrastructure
such as the mail transport agents (MTA),
the directory service agents (DSA),
and the mail stores (MS)
are designed, installed, maintained and operated
by an institutional service provider.
This strategy ensures that the client
and the Mail User Agent (MUA)
view the electronic mail service infrastructure
as a black box that simply ``works''.
Central operations
ensures that the reliability and performance guarantees
necessary for a ``mission critical'' service
that addresses each domain's daily workplace needs
can be met.
The result
of decoupling the client environment
from the technical infrastructure of the service
is a more robust, reliable system.