93.5.10

         Comparing Two Approaches to Remote Mailbox Access: 
                           IMAP vs. POP

                            Terry Gray
               Director, Networks & Distributed Computing
                     University of Washington

There are several different approaches to building a distributed electronic
mail infrastructure.  For example: LAN-oriented, vendor specific systems;
single time-sharing machine solutions; and Internet-oriented
mailserver-based solutions.  The principal options in this last category
are DMSP (Distributed Mail System Protocol), POP (Post Office Protocol),
and IMAP (Interactive Mail Access Protocol).  These protocols are more
robust foundations for a distributed email system than vendor-specific
systems requiring gateways to Internet mail.  Of these three, POP is the
oldest and consequently the best known.  DMSP is largely limited to a
single application, PCMAIL, and is known primarily for its disconnected
(offline) operation capabilities. IMAP offers significant advantages over
POP.  This gap is likely to widen as a result of the imminent addition of
disconnected operation extensions to IMAP. 

With POP (Post Office Protocol), mail is delivered to a shared server, and
a personal computer user periodically connects to the server and downloads
all of the pending mail to the "client" machine.  Thereafter, all mail
processing is local to the client machine.  Think of POP as providing a
store-and-forward service, intended to move mail (on demand) from an
intermediate server (drop point) to a single destination machine, usually a
PC or Mac. Once delivered to the PC or Mac, the messages are typically
deleted from the POP server. 

IMAP is a client-server mail protocol designed to permit manipulation of
remote mailboxes as if they were local.  With IMAP, mail is again delivered
to a shared server, but the mail client machine does not normally copy it
all at once and then delete it from the server.  It's more of a
client-server model, where the IMAP client can ask the server for headers,
or the bodies of specified messages, or to search for messages meeting
certain criteria. Messages in the mail repository can be marked as deleted
and subsequently expunged, but they stay on the repository until the user
takes such action.  Depending on the IMAP client implementation and the
mail architecture desired by a system manager, the user may save messages
directly on the client machine, or save them on the server, or be given
the choice of doing either. 

While POP and IMAP both allow access to mail on a remote server from a
variety of different client platforms, they reflect two different paradigms
and styles of use.  POP works best for people who use a single client
machine all the time; it is not well-suited for the goals of accessing
one's inbox of recent messages or saved-message folders from different
places and different machines at different times. 

The strength of POP, other than its wide availability, is that it minimizes
use of server resources and connect time when used via dialup.  However,
since IMAP is a functional superset of POP, it can also be used in the "POP
paradigm" of connecting to a mail server, retrieving all the pending
messages, and disconnecting.  Thus, the only advantage of the POP
*protocol* over IMAP relates to software availability and not
functionality.  As the amount of IMAP software is growing rapidly, the
historic prevalence of POP is of diminishing importance when compared to
the many advantages of IMAP. 

Because IMAP can mimic all of the POP mail retrieval functions, it is
useful to distinguish the characteristics of the IMAP and POP *paradigms*,
as well as the protocols themselves.  The paradigms define what the user
can do in each model; the protocol characteristics relate to efficiency,
performance, etc.  Here are some of the key similarities and differences
between the two...

o POP and IMAP reflect two different paradigms:
  -POP = store-and-forward (usually to a single client).
  -IMAP = multiple client-server mailbox access.

o Characteristics common to both POP and IMAP:
  -Mail is delivered to a shared, "always up" mail server.
  -New mail accessible from a variety of client platform types.
  -New mail accessible from anywhere in network.
  -Offline mail processing possible, though neither designed for it.
  -Protocols are open; defined by Internet RFCs.
  -Freely available implementations (including source) available.
  -Clients available for PCs, Macs, and Unix.
  -Commercial implementations available.
  -Internet oriented; no SMTP mail gateways required.

o POP paradigm advantages:
  -Minimum use of connect time.
  -Minimum use of server resources.

o POP protocol advantages:
  -Simpler protocol; easier to implement.
  -More client software currently available.

o IMAP paradigm advantages:
  -Saved-message folders may be stored on server (as well as INBOX).
  -Allows access to INBOX (not just new mail) from multiple platforms.
  -Allows concurrent access to a shared mailbox from multiple platforms.
  -Allows concurrent access to mailboxes on multiple mailservers.
  -Offers improved offline mail handling.
  -Allows selective transfer of messages/parts to client (local Save).
  -Can also use POP paradigm, for minimum connect time and server resources.

o IMAP protocol advantages:
  -Suitable for accessing non-email data; e.g., NetNews, documents.
  -Faster startup time, as only the headers are fetched initially.
  -Allows selective fetching of individual MIME message body parts.
  -Effective over low-speed links.
  -Ability to use server for searching.
  -Offline processing w/resynchronizing is a planned enhancement.

"Saved-message folders may be stored on server (as well as INBOX)" allows
"dataless" clients and/or nomadic users (e.g. student labs). 

"Allows access to INBOX (not just new mail) from multiple platforms" means
that if you have a Mac in your office, and PC at home, and a Unix machine
in the lab, you can move freely among them and access the same INBOX. 

"Allows concurrent access to a shared mailbox from multiple platforms."
This capability is useful when multiple individuals are processing messages
coming into a common inbox.  Changes in mailbox state can be presented to
all concurrently active clients via IMAP. 

"Allows concurrent access to multiple inboxes on multiple mailservers."
This is useful for people who have partitioned their incoming mail streams,
either via delivery filters, or by having different accounts for different
purposes. 

IMAP "offers improved offline mail handling" compared to POP. Unlike the
DMSP protocol used in the PCMAIL program, neither POP nor IMAP was designed
with offline use as a primary goal.  However, POP is widely used for this,
even though it is not particularly well-suited for the task. POP requires
you to either entrust all of your mail to your client machine (which may be
about to go thru an airport xray machine), or to over-ride the normal POP
server behavior of deleting the mail on the server, and manually
resynchronizing the diverging mailbox states at a later time.  IMAP can do
better:  you can connect to the server, save to a local folder all or
selected messages, and disconnect.  The advantages over POP are that (1)
the saved messages may be retained on the server, but *marked* as deleted,
so they can be distinguished later from unselected or more recent messages,
and expunged once it is clear they won't be needed, and (2) the ability to
save (download) selectively --especially important when one has a 2MB audio
message in the mailbox and is reading mail via a low-bandwidth connection
from a machine that has no sound capability. 

"Allows selective transfer of messages/parts to client (local Save)."
Especially when connecting to a mail server via low-bandwidth lines, it is
useful to be able to defer transferring messages that are not of immediate
interest until a more propitious time.  Moreover, with multimedia or
multipart MIME messages, transferring selected parts of a message in
increasingly useful. Efficient processing of MIME messages is one of the
major advantages of IMAP over POP.  MIME stands for Multipurpose Internet
Mail Extensions.  It is a technique for encoding arbitrary files as
attachments to SMTP and RFC-822 compatible Internet mail messages.  This is
something that proprietary, LAN-oriented, mail systems have had for some
time, and now is finally available for the Internet.  It allows one to send
spreadsheets, word processing docs, images, and audio to 5 or 10 million of
"your closest Internet friends".  MIME has one particularly nice
capability: it allows inclusion of alternate representations.  For example,
a plain-text version of a document, plus a fax or RTF version.  With IMAP,
the receiving mail user agent gets to decide which message parts to
transfer and present to the user, falling back to plain text if that's all
it can do. 

Even though MIME support is not yet pervasive, its importance and impact on
the IMAP/POP question should not be underestimated. There is tremendous
pent-up demand for this capability and it is taking off fast.  In fact,
people are already using MIME for things that have nothing to do with mail
(e.g. encoding different representations of technical documents, for
campus-wide information systems.) MIME is orthogonal to IMAP and POP,
except that IMAP and MIME are extremely complementary, and there are
already IMAP clients that understand MIME.  (POP clients can and will be
taught to understand MIME, too, but the fact that POP copies all pending
messages at startup, and MIME messages can be very large, means that POP
users may need to become even more patient!)

"Can also use POP paradigm, for minimum connect time and server resources."
The POP paradigm is of interest in situations where the only access to a
mail server is via expensive dialup connections and multi-platform access
to one's inbox(es) is not needed.  It is also useful in environments where
client machines are resource-rich and servers are resource-poor.  However,
because IMAP is a superset of POP functionality, IMAP can be used in "POP
mode". That is, IMAP clients can be designed to provide the option of
transfering all messages to the client and processing them locally
(generally offline), thus providing the same advantages POP has in terms of
minimizing connect time and use of server resources. 

In summary, the fact that IMAP provides access to a persistent remote mail
store, and does not move all pending messages to the client machine, offers
more flexible access to that mail, and provides significant performance
advantages over POP in terms of start-up time and access to large MIME
messages.  The option of accessing remote saved-message folders and/or
NetNews via IMAP (and from the same mail user agent) provides additional
architectural flexibility in comparison to POP. 

A reasonable conclusion is that the only advantage of POP over IMAP is
that there is currently more POP software available.  However, this is
changing rapidly, and IMAP's functional advantages over POP are nothing
less than overwhelming. 

IMAP2 is defined in RFC-1176.  An "IMAP2bis" document describing recent
extensions to IMAP is available (along with various IMAP clients and a
server) in the /mail directory of ftp.cac.washington.edu.  Also available
from the same place is a POP server that, in addition to offering the
normal POP service, can relay commands to an IMAP server, thus permitting
existing POP clients to access an IMAP server.  The IMAP server available
from UW also incorporates a NetNews driver, used at UW as part of its
Campus-Wide Information System. 

---------------------------------------------------------------------------