RFC 1711






Network Working Group                                        J. Houttuin
Request for Comments: 1711                                          RARE
Category: Informational                                     October 1994


                   Classifications in E-mail Routing

Status of this Memo



   This memo provides information for the Internet community.  This memo
   does not specify an Internet standard of any kind.  Distribution of
   this memo is unlimited.

Abstract



   This paper presents a classification for e-mail routing issues. It
   clearly defines commonly used terminology such as static routing,
   store-and-forward routing, source routing and others. Real life
   examples show which routing options are used in existing projects.

   The goal is to define all terminology in one reference paper. This
   will also help relatively new mail system managers to understand the
   issues and make the right choices. The reader is expected to already
   have a solid understanding of general networking terminology.

   In this paper, the word Message Transfer Agent (MTA) is used to
   describe a routing entity, which can be an X.400 MTA, a UNIX mailer,
   or any other piece of software performing mail routing functions. An
   MTA processes the so called envelope information of a message. The
   term User Agent (UA) is used to describe a piece of software
   performing user related mail functions. It processes the contents of
   a message's envelope, i.e., the header fields and body parts.

Table of Contents



   1.   Naming, addressing and routing                               2
   2.   Static versus dynamic                                        4
   3.   Direct versus indirect                                       5
   3.1.       Firewalls                                              5
   3.2.       Default versus rule based                              6
   4.   Routing at user level                                        7
   4.1.       Distributed domains                                    7
   4.2.       Shared MTA                                             8
   5.   Routing control                                              9
   6.   Bulk routing                                                 9
   7.   Source routing                                              11
   8.   Poor man's routing                                          12
   9.   Routing communities                                         12



Houttuin                                                        [Page 1]

RFC 1711           Classifications in E-mail Routing        October 1994


   10.  Realisations                                                14
   10.1.      Internet mail                                         14
   10.2.      UUCP                                                  15
   10.3.      EARN                                                  15
   10.4.      GO-MHS                                                15
   10.5.      ADMD infrastructure                                   15
   10.6.      Long Bud                                              16
   10.7.      X42D                                                  16
   11.  Conclusion                                                  16
   12.  Abbreviations                                               17
   13.  References                                                  17
   14.  Security Considerations                                     19
   15.  Author's Address                                            19

1.    Naming, addressing and routing



   A name uniquely identifies a network object (without loss of
   generality, we will assume the 'object' is a person).

   Once a person's name is known, it can be used as a key to determine
   his address.

   An address uniquely defines where the person is located. It can
   normally be divided into a domain related part (e.g., the RFC 822
   domainpart or in X.400 an ADMD or OU attribute) and a local or user
   related part (e.g., the RFC 822 localpart or in X.400 a DDA or
   Surname attribute). The domain related part of an address typically
   consists of several components, which normally have a certain
   hierarchical order. These domain levels can be used for routing
   purposes, as we will see later.

   Once a person's address is known, it can be used as a key to
   determine a route to that person's location.

   We will use the following definition of an e-mail route:

       e-mail route           a path between two leaves in a
                              directed Message Transfer System
                              (MTS) graph that a message travels
                              for one originator/recipient pair.
                              (see Figure 1)

   Note that, in this definition, the User Agents (UAs) are not part of
   the route themselves. Thus if a message is redirected at the UA
   level, a new route is established from the redirecting UA to the UA
   the message is redirected to.





Houttuin                                                        [Page 2]

RFC 1711           Classifications in E-mail Routing        October 1994


   The first and last leaves in a mail route are not always UAs. A route
   may start from a UA, but stop at a certain point because one of the
   MTAs is unable to take any further routing decisions. If this
   happens, a warning is generated by the MTA (not by a UA), and sent
   back to the originator of the undeliverable message. It may even
   happen that none of the leaves is a UA, for instance if a warning
   message as discussed above turns out to be undeliverable itself. The
   cautious reader may have noticed that this is a dangerous situation.
   Special precautions are needed to avoid loops in such cases (see
   [1]).

           user                        user
            |                           ^
            v                           |
     +-----------------------------------------+
     |      |                           ^      |
     |      v                           |      |
     |   +-----+                     +-----+   |
     |   | UA  |                     | UA  |   |
     |   +-----+                     +-----+   |
     |      |                           ^      |
     |      v                           |      |
     | +-------------------------------------+ |
     | |    v                           ^    | |
     | |    v                           ^    | |
     | |    v                           ^    | |
     | | +-----+                     +-----+ | |
     | | | MTA |.....................| MTA | | |
     | | +-----+                     +-----+ | |
     | |    v   \                       ^    | |
     | |    v    \................      ^    | |
     | |    v                     \     ^    | | NB The actual route
     | | +-----+                   \ +-----+ | |    is drawn as
     | | | MTA |>>>>>>>>>>>>>>>>>>>>>| MTA | | |    v            ^
     | | +-----+                     +-----+ | |    v            ^
     | | Message Transfer System             | |    v  >>>>>>>>  ^
     | +-------------------------------------+ |
     | Message Handling System                 |
     +-----------------------------------------+

                Figure 1. A mail route

   It is important that the graph is directed, because routes are not
   necessarily symmetric. A reply to a message may be sent over a
   completely different mail route for reasons such as cost, non-
   symmetric network connectivity, network load, etc.





Houttuin                                                        [Page 3]

RFC 1711           Classifications in E-mail Routing        October 1994


   According to the definition, if a message has two different
   recipients, there will also be two mail routes. Since the delivery to
   a UA (not the UA itself) is a part of the route, this definition is
   still valid if two UAs are connected to the same MTA.

   The words '.. for one originator-recipient pair.' in the definition
   do not imply that this pair provides the MTA with all necessary
   information to choose one specific route. One originator-recipient
   pair may give an MTA the possibility to choose from a number of
   possible routes, the so-called routing indicators (see chapter 2).

   Other information (e.g., line load, cost, availability) can then be
   used to choose one route from the routing indicators.

   Routing is defined as the process of establishing routes. Note that
   this is a distributed process; every intermediate MTA takes its own
   routing decisions, thus contributing to the establishment of the
   complete route.

   Taking a routing decision is not a purely algorithmic process,
   otherwise there would hardly be any difference between an address and
   a route. The address is used as a key to find a route, typically in
   some sort of rule-based routing database. The possible options for
   realising this database and algorithms for using it are the subject
   of the rest of this paper.

2.    Static versus dynamic



   Dynamic (mail) routing allows a routing decision to be influenced by
   external factors, such as system availability, network load, etc. In
   contrast, static (mail) routing is not able to adapt to environmental
   constraints. Static routing can be viewed as an extremely simple form
   of dynamic routing, namely where there is only one choice for every
   routing decision.

   Dynamic routing algorithms normally use some kind of distributed
   databases to store and retrieve routing information, whereas static
   routing is typically implemented in routing tables.

   Note that dynamic routing can occur at different layers: at the mail
   level, dynamic routing might allow a message to be relayed to a
   choice of MTAs (the routing indicators). As an example, consider the
   Internet mechanism of using multiple Mail eXchanger (MX) records,
   describing MTAs that can serve a domain. If the primary choice MTA is
   not available, a second choice MTA can be tried. If this second
   choice MTA is busy, a third one will be tried, etc. On lower layers,
   there may be more than one presentation address for one MTA, each of
   which can again have an associated priority and other attributes.



Houttuin                                                        [Page 4]

RFC 1711           Classifications in E-mail Routing        October 1994


   These choices may represent that an MTA prefers to be connected to
   using one certain stack, e.g., RFC1006/TCP/IP, but is also able to
   accept incoming calls over another stack, such as TP0/CONS/X.25. We
   will call this dynamic stack routing. Theoretically, dynamic stack
   routing should be transparent to the mail routing application, and is
   thus not a part of dynamic mail routing. It is mentioned here because
   in existing products, dynamic stack routing is often very well
   visible at the mail configuration level, so MTA managers should at
   least be aware of it.

   Although static routing is often table based, not all table based
   routing algorithms are necessarily static in nature. As a counter
   example, X.400 routing according to RFC 1465 [2] is clearly table
   based, but at the same time allows a fairly dynamic kind of mail
   routing (as well as dynamic stack routing, which in this approach is
   cleanly separated from the dynamic mail routing part). A mail domain
   can specify a choice of so-called RELAY-MTAs (formerly called WEPs)
   that will serve it, each with a priority and maximum number of
   retries.

   For reasons of flexibility and reliability, dynamic routing is almost
   always the preferred method.

3.    Direct versus indirect



   Direct routing allows the originator's MTA to contact the recipient's
   MTA directly, whereas indirect routing (also known as store-and-
   forward routing) uses intermediate MTAs to relay the message towards
   the recipient. It is difficult to clearly distinguish between direct
   and indirect routing: direct routing assumes the existence of a fully
   meshed routing topology, whereas indirect routing assumes the
   existence of a more tree-like hierarchical topology. Mail routing in
   most existing networks is upto some degree indirect. Networks can be
   classified as being more or less direct according for the following
   rule of thumb: larger fan out of the routing tree means more direct
   routing, greater depth of the tree means less direct routing. Two
   kinds of indirect routing are presented here: firewalls (downstream)
   and default routes (upstream).

3.1.  Firewalls



   A firewall 'attracts' all messages for a certain set of addresses
   (the address sub space behind the firewall) from the outside e-mail
   world to a central relaying MTA (the firewall). This is done by
   publishing routes to all other MTAs that must relay their messages
   over this firewall (the attracted community). Note that local
   knowledge should be used to route messages within the address space
   behind the firewall. An example for this is presented later on. There



Houttuin                                                        [Page 5]

RFC 1711           Classifications in E-mail Routing        October 1994


   exist many reasons for using firewalls, e.g., security considerations
   or to concentrate the management for a given domain onto one well
   managed system.

   The Internet mail system would allow all mail hosts connected to the
   Internet to directly accept mail from any other host, but not all
   hosts use this possibility. Many domains are hidden behind one or
   more 'Mail eXchanger' (MX), which offer to relay all incoming mail
   for those domains. The RELAY-MTAs mentioned earlier can also be
   considered firewall systems.

         +-----------------------------------+
         |                                   |
         | The rest of the e-mail world      |
         |                                   |
         +-----------------------------------+
                     \  |  |   /
                      \ |  |  /
                       \|  | /
                        v  vv
                  +--------------+
                  |Firewall MTA A|
                  +--------------+
                    ^  /  ^  \  ^
                   /  /   |   \  \
                  /  /    |    \  \
  Default route--o  /     |     \  o---Default route
                /  /      |      \  \
               /  /       |       \  \
              /  v        v        v  \
           +-----+     +-----+   +-------+
           |MTA B|<----|MTA C|   |MTA D  |
           +-----+     +-----+   +-------+
            /  |         |         |   \
           /   |         |         |    \
          /    |         |         |     \
       +----+ +----+  +----+   +----+ +----+
       | UA | | UA |  | UA |   | UA | | UA |
       +----+ +----+  +----+   +----+ +----+

        Figure 2. Firewall and default route

3.2.  Default versus rule based



   Default routing is to outgoing mail what a firewall is for incoming
   mail, and is thus often used in conjunction with firewalls. It is
   about the simplest routing algorithm one can think of: route every
   message to one and the same MTA, which is trusted to take further



Houttuin                                                        [Page 6]

RFC 1711           Classifications in E-mail Routing        October 1994


   care of routing the message towards its destination. Pure default
   routing is rather useless; it is normally used as a fall back
   mechanism accompanying a rule based algorithm.

   For example, the simplest usable default algorithm is the following:
   first check if a mail should be delivered to a local UA. If not,
   perform default routing.

   In order to avoid loops, it is not acceptable for all MTAs within a
   certain routing community (see chapter 9) to use default routing. At
   least one MTA should be able to access all routing rules for that
   community. Consider the following example: An X.400 MTA A, which
   serves the organisation organisational unit OU=orgunA within the
   organisation O=org, receives a message for the domain O=org;
   OU=orgunB;. Since MTA B in the same organisation serves all other
   OUs, A will default route the message to B. Suppose that B would use
   the same mechanism: first check if the OU is local and if not,
   default route to A. If OU=orgunC is not served by either A or B, this
   routing set-up will lead to a loop. The decision that a certain OU
   does not exist can only be made if at least one of the MTAs has
   knowledge of all existing OUs under O.

   An example of a firewall and two default routes is shown in figure 2.
   It visualises that a firewall is a downstream and a default route is
   an upstream indirection. MTA B and D use default routing; they can
   only route to one other MTA, MTA A.

   For more detailed information, please refer to [3], which lists most
   pros and cons of both approaches. Your choice will depend on many
   factors that are specific for your messaging environment.

4.    Routing at user level



   Normally a message is routed down to the deepest level domain, and
   then delivered to the recipient per default routing. I.e., every user
   in this domain is considered to have his mailbox uniquely defined
   within this domain on the same MTA, and every user on that MTA can be
   distinguished within this domain. Exceptions can occur when the users
   within a domain have their mailboxes on different MTAs (distributed
   domain), or when several domains exist on the same MTA (shared MTA).

4.1.  Distributed domains



   Routing is normally performed down to a certain domain level. Mail to
   all users that are directly registered under this domain is then
   delivered per default routing, i.e., delivered locally. Explicit user
   routing (i.e., rule-based routing on user level attributes according
   to a fixed table listing all users) may be necessary when not all



Houttuin                                                        [Page 7]

RFC 1711           Classifications in E-mail Routing        October 1994


   users have their UAs connected to the same MTA.

   Note that the whole issue of distributed domains is nothing more than
   a special case of the problems discussed in chapter 3.2: 'Default
   versus rule-based'. The only reason for mentioning this in a separate
   chapter is that there are many software products that don't deal with
   routing based on local address parts in the same way as with routing
   based on domain related address parts.

   As an example, consider an organisation where two mail platforms are
   available. Some users prefer using platform A, others prefer platform
   B. Of course, the easiest solution would be to create a subdomain A
   and a subdomain B, and then route domain A to system A and B to B.
   Default user routing on both platforms would then do the rest.
   However, when an organisation wants to present itself to the outside
   world using only one domain, this scheme cannot be used, at least not
   without special precautions (see the paragraph about avoiding loops
   in chapter 3.2).

     +----------+      +---------------------------+
     |   MTA A  |      |        Shared MTA B       |
     +----------+      +---------------------------+
        |     |         /        |     |        |
     +-----------------/----+ +-----------+  +----------+
     |  |     |       /     | |  |     |  |  |  |       |
     | +--+ +--+ +--+/      | | +--+ +--+ |  | +--+     |
     | |UA| |UA| |UA|       | | |UA| |UA| |  | |UA|     |
     | +--+ +--+ +--+       | | +--+ +--+ |  | +--+     |
     | Distributed Domain A | | Domain B  |  | Domain C |
     +----------------------+ +-----------+  +----------+

   Figure 3. Distributed domains and shared MTAs

   Another possibility to have uniform addresses in outgoing e-mail,
   despite the fact that a domain is distributed, is to make routing
   decisions on information in the local part of the address, e.g., in
   X.400 the Surname in exactly the same manner as making routing
   decisions on any other attributes. Thus products and routing
   algorithms that are able to route on user related address parts are
   said to support distributed domains.

4.2.  Shared MTA



   The opposite of a distributed domain is a shared MTA: several domains
   are routed locally on the same MTA. These domains sharing one MTA may
   cause problems when two or more domains have a local user with the
   same name.




Houttuin                                                        [Page 8]

RFC 1711           Classifications in E-mail Routing        October 1994


   Theoretically, this problem doesn't exist: the address is being
   routed down to the deepest domain level, and within that level, there
   will only be one user with that name (let's at least assume this for
   simplicity). Some products however use only one database of all users
   locally connected to this MTA instead of one database per domain, so
   that default user routing at the deepest level can lead to conflicts.
   It is beyond the scope of this document to describe the tricks needed
   to avoid these conflicts when using such products.

5.    Routing control



   Routing control means that routing decisions can be affected by the
   originator of a message. This normally takes the form of either
   granting or denying access for a certain user or group of users.

   Routing control is often useful in an X.400 ADMD/PRMD environment,
   where it is either used to grant access only to users who are known
   to be chargeable, or where ADMDs can refuse messages that were
   relayed to them over international PRMD connections; a policy that is
   not allowed in the CCITT version of the standards (as opposed to the
   ISO version). Of course, the PRMDs can also perform routing control
   themselves in order to circumvent such problems.

   Although there may be good reasons for using routing control, one
   must be aware that it can make the messaging environment
   unpredictable for end-users. Where using routing control is
   unavoidable, the originator whose message has been rejected is likely
   to appreciate receiving a message, clearly telling him where and why
   routing of his message was refused, whom to contact, and what options
   are available to avoid such rejections in the future.

6.    Bulk routing



   In order to reduce network traffic, intelligent mailers may prefer a
   message addressed to a group of remote users to be transferred to a
   remote domain only once, thus postponing the 'explosion' into several
   copies. This technique, called bulk routing, is especially useful
   when an MTA hosts large mailing lists.

   Several possibilities exist. In a typical hierarchical firewall mail
   system, bulk routing can be done almost automatically by intelligent
   MTAs. For instance, in an X.400 community, a large international
   distribution list can create a message with an envelope containing
   1000 recipient addresses, some of which can probably be grouped by
   the MTA depending on whether they can be routed further to the same
   remote MTA, according to the normal routing implementation at this
   MTA. The size and number of these groups will largely depend on how
   indirect this routing implementation is. In the GO-MHS community, the



Houttuin                                                        [Page 9]

RFC 1711           Classifications in E-mail Routing        October 1994


   number of groups will almost always be less than 50, which provides a
   rather fair distribution of traffic load over the involved MTAs (that
   is, fair according to the author's taste, who is not aware of any
   existing fair mail load distribution formula).

   As an extreme example, the simplest way to automatically (i.e.,
   without using special optimisation tools) bulk route mail is to use
   one default route. Any outgoing message, regardless of the number of
   recipients, will be routed over the default route only once. The
   default remote MTA will then have to break up the message (envelope)
   into several copies and is thus responsible for the actual explosion
   and distribution. NB. This mechanism can be exploited to shift the
   cost and overhead of exploding a message towards another domain/MTA.
   If you ever get a request for a bilateral default route agreement;
   i.e., the requesting party wants to default route over your MTA, it
   may be worth to check first if the requesting party is running or
   planning to run large mailing lists.

   In more direct routing environments, such as BITNET, bulk routing
   will not function as automatically as described above. Without
   special precautions, an MTA would open a direct connection to every
   single host that occurs in the message's envelope, regardless of
   whether some of these hosts are far away from this MTA, but close to
   each other, measured by underlying network topology. This can clearly
   lead to a waste of expensive bandwidth. In order to be able to detect
   such cases, and to act upon it by sending one single copy over an
   expensive link and have it distributed at some remote hosts, an MTA
   must have additional knowledge of the relation between mail domains
   and the underlying network topology.

   BITNET uses the distribute protocol [4] for this purpose. A selected
   set of hosts is published to have the required topology knowledge and
   to be able to efficiently distribute the mail on behalf of other
   MTAs, who can explicitly route all bulk mail to one of those hosts.
   The complete message, including the envelope, is encoded in a message
   body, which starts with a distribution request to the distribute
   server. This server will break up the rest of the body into the
   original envelope and contents and then use it's topology knowledge
   to efficiently distribute the original message. Note that this
   protocol violates the conceptual model of the layering of MTA and UA
   functionality, but it is about the only trick that will work in a
   very direct routing environment. It is only needed to overrule a non-
   efficient (for large mailing lists) routing topology.

   Bulk routing is an area where mail routing issues start to overlap
   with the area of distributing netnews (bulletin board services).
   Several organisations, such as ISO, RARE and the IETF have started
   initiatives in the direction of harmonising the two worlds. The first



Houttuin                                                       [Page 10]

RFC 1711           Classifications in E-mail Routing        October 1994


   results, be it standards or products, are not expected before 1995
   though.

7.    Source routing



   Source routing was originally intended to allow a user to force a
   message to take a certain route. The mechanism works as follows: the
   MTA that the user wants the message to be routed through is
   integrated into the address. Once the message has arrived at the
   specified MTA, that MTA strips itself from the source-routed address
   and routes the remaining address in the usual way. This mechanism is
   called explicit source routing and can be useful if a user wants to
   test a routing path or force a message to be routed over a faster,
   cheaper, more reliable, or otherwise preferred path.

   For instance, if the Internet user user@uni-a.edu wants to test the
   mail connections to and from a remote domain uni-b.edu, he might
   source route a message to himself over the MTA at uni-b.edu by
   addressing the mail to:  @uni-b.edu:user@uni-a.edu

   Source routing need not always be explicit. Source routes can also be
   generated automatically by a gateway, in which case we speak of
   address rooting (to that gateway). The gateway will root itself to
   the message by putting its own domain in the source route mapped
   address, thus ensuring that any replies to the gatewayed message will
   be routed back through the same gateway.

   Example 1: RFC 1327 left hand side encoding (see [5]) performed by
   the gateway 'gw.ch':

        C=zz;A=a;P=p;O=oo;S=plork ->
        "/C=zz/A=a/P=p/O=oo/S=plork/"@gw.ch

   Example 2: RFC 1327 DDA mapping (see [5]) performed by the gateway
   C=zz;A=a;

        bush@dole.us ->
        DD.RFC-822=bush(a)dole.us;C=zz;A=a;

   Example 3: the so-called %-hack:

        user%final.domain@1st.relay

   When the relaying host '1st.relay' receives the message, it strips
   its own domain part and interprets the localpart 'user%final.domain':
   it changes the % to an @ sign and relays the message to the address

        user@final.domain



Houttuin                                                       [Page 11]

RFC 1711           Classifications in E-mail Routing        October 1994



   Example 4: Another example of the already mentioned explicit source
   routing, this time through two relays:

        @1st.relay,@2nd.relay:user@final.domain

   In the Internet, use of explicit source routing is strongly
   discouraged (see [6]), one reason being that not all mail relays will
   handle such addresses in a consistent manner. Apart from that, the
   need to use explicit source routing has disappeared over the last
   decennia. In earlier days, when the RFC 822 world consisted of many
   sparsely connected 'mail islands', source routing was sometimes
   needed to make sure that a message was routed through a gateway that
   was known to be connected to a remote island. Nowadays, the RFC 822
   world is almost fully interconnected through the Internet, so the
   need for end-users to have knowledge of the mail network's topology
   has become superfluous.

8.    Poor man's routing



   If we combine static, indirect and source routing, we get what is
   commonly known as "poor man's routing". The user thus specifies the
   complete route in the address. A classic example is the old UUCP bang
   style addressing:

        host1!host2!host3!host4!user

   Poor man's routing is presented here for historical reasons only.
   Since, for reasons discussed earlier, most present networks
   discourage source routing and prefer dynamic over static routing,
   poor man's routing is not widely deployed anymore.

9.    Routing communities



   A routing community can be defined as follows:

       Routing community:     a set of MTAs X, with the property
                              that for any address a, every MTA
                              in X except a subset Ya will have
                              the option, according to an agreed
                              upon set of routing rules, to
                              directly route that address to at
                              least one MTA in Ya.

   Which is a rather formal way of describing that a routing community
   consists of a set of MTAs (and human operators) that agreed on a
   common set of rules on how to route messages among each other.




Houttuin                                                       [Page 12]

RFC 1711           Classifications in E-mail Routing        October 1994


   An example of a routing community is the large Internet routing
   community, in which the agreed rules are implemented in the Domain
   Name System (DNS). For details, refer to [7]. The subset Ya is in
   this case the set of MTAs that have an MX record in the DNS for a.
   MTAs that hide behind fire walls or behind default routes are thus
   not considered direct members of this community, but normally form
   their own smaller routing community, with one host (the mail
   exchanger/default route) belonging to both communities.

   Another example is the GO-MHS community, consisting of a set of
   documented RELAY-MTAs (formerly called WEPs, Well-known Entry
   Points). Routing communities can be further classified depending on
   the openness and topology of their routing rules. [3] defines four
   classes of routing communities:

       Local community:       The scope of a single MTA. Contains
                              the MTAs view of the set of
                              bilateral routing agreements, and
                              routing information local to the
                              MTA. Example: any local MTA.

       Closed community:      This is like a local community, but
                              involves more than one MTA. The
                              idea is to route messages only
                              within this closed community. A
                              small subset of the involved MTAs
                              can be in another community as
                              well, in order to get the
                              connectivity to the outside world,
                              as described earlier. Example: A
                              set of Private Management Domains
                              (PRMDs) representing the same
                              organisation in multiple countries.

       Open community:        All routing information is public
                              and any MTA is invited to use it.
                              Example: the Internet.

       Hierarchical community:A subtree of the O/R address tree.
                              Note that the subtree will in
                              practice often be pruned; sub-sub-
                              trees may form their own routing
                              community. Example: GO-MHS.

   This classification cannot always be followed too strictly. For
   example, completely closed communities are relatively rare. In order
   for e-mail to be an effective communication tool, an organisation
   will typically designate at least one of its MTAs as a gateway to



Houttuin                                                       [Page 13]

RFC 1711           Classifications in E-mail Routing        October 1994


   another routing community, for instance to the Internet. The
   organisation will register an Internet domain, say 'org.net', which
   points to this gateway, and thus acts as a firewall from the Internet
   to the domain 'org.net', and as a default route from the closed
   community to the rest of the Internet. At this stage, the gateway MTA
   can be regarded as being a member of any of the four types of routing
   communities. The reader is invited to check this himself.

   Especially the distinction between open and closed communities is not
   always easy. To some extent, most routing communities are open, at
   least among their own participants. It is just that some routing
   communities are more open than others. Also, even the most open
   routing community is not just open to anyone. It is not enough for a
   community participant to use the community's routing rules and
   connect to any other MTA in the community. The participant will
   typically also have to fulfil an agreed upon set of operational
   requirements, for example the Internet host requirements [6] or the
   GO-MHS domain requirements [8].

   The most open routing community known today is certainly the Internet
   mail community. As for X.400 routing communities, some problems occur
   when trying to open a community, the main one being that most X.400
   software does not support the so called 'anonymous' connection mode,
   which allows any remote MTA to connect to it. Most software was
   designed or configured to use passwords for setting up MTA
   connections. This, together with the fact that X.400 routing was
   originally designed to be hierarchical, is one of the main reasons
   why most X.400 communities today are either closed or hierarchical.

10.   Realisations



   In this chapter some of the routing classifications described above
   are assigned to existing mail services and projects.

10.1. Internet mail



   RFC 822 mail. An operational service. Co-ordination: distributed.
   Mostly dynamic routing, although static routing is also possible. DNS
   based routing rules(*). Mostly direct routing, although indirect is
   also possible. No dynamic stack routing. Distributed domains
   possible. Shared MTAs possible, but rare. Routing control not
   normally used. Bulk routing via SMTP envelope grouping; also
   possible, but not widely deployed, using the 'distribute protocol'
   [4]. Source routing supported, but strongly discouraged. No poor
   man's routing. Open (and hierarchical) routing community.

   (*) Sub-communities don't use DNS based routing: The MX records in
   the DNS are used to "attract" messages from the Internet to the



Houttuin                                                       [Page 14]

RFC 1711           Classifications in E-mail Routing        October 1994


   "border" between the Internet and the sub-community. Thus from the
   Internet we have dynamic, directory based routing but once the
   "border" is reached, it is no longer possible to use MX records for
   mail routing, and thus some form of static routing is generally
   needed.

10.2. UUCP



   RFC 822 style mail. An operational service. Co-ordination:
   distributed. Mostly static routing, although dynamic routing is also
   possible. Table based routing rules. Mostly indirect routing. No
   dynamic stack routing. No distributed domains. Shared MTAs possible,
   but rare. Routing control not normally used. No bulk routing
   possible. Source routing (poor man's routing) still widely used by
   means of 'bang' addressing, but strongly discouraged. Open (and
   hierarchical) routing community.

10.3. EARN



   BITNET mail. An operational service. Co-ordination: The EARN Office,
   France. Static routing. Table based routing rules, although an X.500
   based experiment is running. Mostly direct routing, although indirect
   is also possible. No dynamic stack routing. No distributed domains.
   No shared MTAs. Routing control not normally used. Bulk routing
   possible using the 'distribute protocol' [4]. Source routing not
   supported. No poor man's routing. Open routing community.

10.4. GO-MHS



   X.400 mail. An operational service. Co-ordination: GO-MHS Project
   Team, Switzerland. Mostly static routing, although dynamic routing is
   getting more and more deployed since the introduction of RFC 1465
   [2]. Table based community-wide routing rules. Indirect routing.

   Dynamic stack routing. Distributed domains possible. Shared MTAs.
   Routing control not normally used, only to avoid routing control
   problems when routing international traffic to ADMDs. Bulk routing
   using X.400 'responsibility' envelope flags. Source routing supported
   for gatewaying to the Internet. No poor man's routing. Hierarchical,
   but open, routing community.

10.5. ADMD infrastructure



   X.400 mail. An operational service. Co-ordination: The joint
   Administrative Management Domains (ADMDs), typically operated by
   PTTs. Mostly static routing. Indirect routing. Table based bilateral
   routing rules. No dynamic stack routing. Distributed domains not
   supported. Shared MTAs. Routing control used to prohibit routing of



Houttuin                                                       [Page 15]

RFC 1711           Classifications in E-mail Routing        October 1994


   international traffic through PRMDs and to limit access to certain
   gateways. Bulk routing using X.400 'responsibility' envelope flags.
   Source routing possible for gatewaying to the Internet. No poor man's
   routing. Closed hierarchical routing community.

10.6. Long Bud



   X.400 mail. A pilot project. Co-ordination: The IETF MHS-DS working
   group. Dynamic routing. X.500 based routing rules. Mostly indirect
   routing, although direct is also possible. Dynamic stack routing.
   Distributed domains. Shared MTAs. No routing control. Bulk routing
   using X.400 'responsibility' envelope flags. Source routing supported
   for gatewaying to the Internet. No poor man's routing. Open
   hierarchical routing community.

10.7. X42D



   X.400 mail. An experiment. Co-ordination: INFN, Italy. Dynamic
   routing. DNS based routing rules as defined in [9]. Mostly indirect
   routing, although direct is also possible. Dynamic stack routing. No
   distributed domains. Shared MTAs. No routing control. Bulk routing
   using X.400 'responsibility' envelope flags. Source routing supported
   for gatewaying to the Internet. No poor man's routing. Open
   hierarchical routing community.

11.   Conclusion



   We have seen several dimensions in which mail routing can be
   classified. There are many more issues that were not discussed here,
   such as how exactly the routing databases are implemented, which
   algorithms to use for making the actual choices in dynamic routing,
   etc. A follow-up paper is planned to discuss such aspects in more
   detail.

   So far, the author has tried to keep this paper free of opinion, but
   he would like to conclude by listing his own favourite routing
   options (without any further explanation or justification; please
   feel free to disagree):

       Static/dynamic:        Dynamic
       Direct/indirect:       Every routing community has its own
                              optimum level of indirection
       User routing:          Support
       Routing control:       Avoid
       Bulk routing:          Efficient distribution should be
                              transparent at mail level, but we
                              may need better e-mail models
                              before this becomes possible



Houttuin                                                       [Page 16]

RFC 1711           Classifications in E-mail Routing        October 1994


       Source routing:        Avoid where possible
       Poor man's routing:    Avoid

12.   Abbreviations



    ADMD              Administration Management Domain
    CCITT             Comite Consultatif International de
                       Telegraphique et Telephonique
    CONS              Connection Oriented Network Service
    DDA               Domain Defined Attribute
    DNS               Domain Name System
    GO-MHS            Global Open MHS
    IP                Internet Protocol
    ISO               International Organisation for Standardisation
    Long Bud          Not an abbreviation
    MHS               Message Handling System
    MHS-DS            MHS and Directory Services
    MTA               Message Transfer Agent
    MTS               Message Transfer System
    MX                Mail eXchanger
    O/R address       Originator/Recipient address
    PP                Not an abbreviation
    PRMD              Private Management Domain
    RARE              Reseaux Associes pour la Recherche Europeenne
    RFC               Internet Request for Comments
    RTR               RARE Technical Report
    SMTP              Simple Mail Transfer Protocol
    STD               Internet Standard RFC
    TCP               Transfer Control Protocol
    TP0               Transport Protocol Class 0
    UA                User Agent
    UUCP              UNIX to UNIX CoPy
    WEP               Well-known Entry Point

13.   References



      [1]         Houttuin, J., "C-BoMBS : A Classification of Breeds
                  Of Mail Based Servers", RARE WG-MSG Work in Progress,
                  April 1994.

      [2]         Eppenberger, E., "Routing Coordination for X.400 MHS
                  Services Within a Multi Protocol / Multi Network
                  Environment Table Format V3 for Static Routing",
                  RFC 1465, SWITCH, May 1993.

      [3]         Kille, S., "MHS use of the Directory to support MHS
                  routing", Work in Progress, July 1993.




Houttuin                                                       [Page 17]

RFC 1711           Classifications in E-mail Routing        October 1994


      [4]         Thomas, E., "Listserv Distribute Protocol",
                  RFC 1429, Swedish University Network, February 1993.

      [5]         Kille, S., "Mapping between X.400(1988) / ISO 10021
                  and RFC 822", RFC 1327, RARE RTR 2, University
                  College London, May 1992.

      [6]         Braden, R., Editor, "Requirements for Internet Hosts
                  - Application and Support", STD 3, RFC 1123, USC/
                  Information Sciences Institute,  October 1989.

      [7]         Partridge, C., "Mail Routing and the Domain System",
                  STD 14, RFC 974, BBN, January 1986.

      [8]         Hansen, A. and R. Hagens, "Operational Requirements
                  for X.400 Management Domains in the GO-MHS
                  Community", Work in Progress, March 1993.

      [9]         Allocchio, C., Bonito, A., Cole, B., Giordano, S.,
                  and R. Hagens "Using the Internet DNS to Distribute
                  RFC1327 Mail Address Mapping Tables", RFC 1664,
                  GARR-Italy, Cisco Systems Inc, Centro Svizzero
                  Calcolo Scientific, Advanced Network & Services,
                  February 1993.

      [10]        Houttuin, J., "A Tutorial on Gatewaying between X.400
                  and Internet Mail", RFC 1506, RTR 6, RARE Secretariat,
                  August 1993.

      [11]        Postel, J., "Simple Mail Transfer Protocol", STD 10,
                  RFC 821, USC/Information Sciences Institute, August
                  1982.

      [12]        Crocker, D., "Standard for the Format of ARPA
                  Internet Text Messages", STD 11, RFC 822, UDEL,
                  August 1982.

      [13]        Alvestrand, H.T., et al, "Introducing Project Long
                  Bud Internet Pilot Project for the Deployment of
                  X.500 Directory Information in Support of X.400
                  Routing", Work in Progress, June 1993.

      [14]        Kille, S., "A Simple Profile for MHS use of
                  Directory", Work in Progress, July 1993.

      [15]        Kille, S., "MHS use of the Directory to Support
                  Distribution Lists", Work in Progress, November 1992.




Houttuin                                                       [Page 18]

RFC 1711           Classifications in E-mail Routing        October 1994


      [16]        Eppenberger, U., "X.500 directory service usage for
                  X.400 e-mail", Computer Networks for Research in
                  Europe No.1: Computer Networks and ISDN Systems 25,
                  Suppl.1 (1993) S3-8, September 1993.

      [17]        CCITT Recommendations X.400 - X.430. Data
                  Communication Networks: Message Handling Systems.
                  CCITT Red Book, Vol. VIII - Fasc. VIII.7, Malaga-
                  Torremolinos 1984.

      [18]        CCITT Recommendations X.400 - X.420. Data
                  Communication Networks: Message Handling Systems.
                  CCITT Blue Book, Vol. VIII - Fasc. VIII.7, Melbourne
                  1988.

14.   Security Considerations



   Security issues are discussed in section 3.1.

15.   Author's Address



   Jeroen Houttuin
   RARE Secretariat
   Singel 466-468
   NL-1017 AW Amsterdam
   The Netherlands

   Phone: +31 20 639 11 31
   Fax:  +31 20 639 32 89
   EMail: houttuin@rare.nl





















Houttuin                                                       [Page 19]