RFC 1022 (rfc1022) - Page 2 of 12

High-level Entity Management Protocol (HEMP)

RFC 1022                     HEMS Protocol                  October 1987


   Two types of interaction are envisioned: a message exchange between
   an application and an entity managed by the application, and
   unsolicited messages from an entity to the management centers
   responsible for managing it.

   When an application wants to retrieve information from an entity or
   gives instructions to an entity, it sends a request message to the
   entity.  The entity replies with a response, either a reply message
   if the request was valid, or an error message if the request was
   invalid (e.g., failed authentication).  It is expected that there
   will only be one response to a request message, although the protocol
   does not preclude multiple responses to a single request.

   Protocol error messages are generated if errors are found when
   processing the HEMP encapsulation of the message.  The possible
   protocol error messages are described later in this document.  Non-
   HEMP errors (e.g., errors that occur during the processing of the
   contents of the message) are application errors.  The existence of
   application error messages does not preclude the possibility that a
   reply will have an error message in it.  It is expected that the
   processing agent on the entity may have already started sending a
   reply message before an error in a request message is discovered.  As
   a result, application errors found during processing may show up in
   the reply message instead of a separate application error message.

   Note that in certain situations, such as on secure networks,
   returning error messages may be considered undesirable.  As a result,
   entities are not required to send error messages, although on
   "friendly" networks the use of error messages is encouraged.

   Event messages are unsolicited notices sent by an entity to an
   address, which is expected to correspond to one or more management
   centers.  (Note that a single address may correspond to a multicast
   address, and thus reach multiple hosts.)  Event messages are
   typically used to allow entities to alert management centers of
   important changes in their state (for example, when an interface goes
   down or the entity runs out of network buffers).














Partridge & Trewitt