Sounds like enough to get started to me! For the face to face?
I would add that we should really investigate S/MIME (CMS)
as an alternative to XML encryption for "bulk" (payload?)
xml encryption. I am wondering whether XML encryption
should be mainly reserved for "smallish" items within
XML documents (passwords, pins, credit card numbers and
small risky items like that); this worry is based
on the base64 expansion issue
that you are not wild about, and that
occurs for basically any xmlencryption process,
as far as I can tell.
I saw that the XMill compression approach is largely
superceded from performance standpoint-- thanks for the reference, Matt.
Dale Moberg
-----Original Message-----
From: Matthew MacKenzie [mailto:]
Sent: Friday, January 31, 2003 10:56 AM
To:
Subject: [ebxml-msg] Compression/Performance Roundup
During the last meeting, we discussed the topic of compression of ebXML
Messages. The discussion branched out into a dialogue on performance
of message transmission and processing.
From that dialogue, I've compiled a summary of the issues that we need
to discuss for version 3.0. These issues are really just talking
points for the general topic of performance, which a v3 implementation
of nearly anything has to take into account.
- Compression algorithms for use with MIME "Transfer Encoding".
- XML Compression
- Compression and its relationship to encryption and digital signature.
- Compression before encryption allegedly provides more
throughput at
the expense of CPU. Do we spec comp-before-enc, or investigate S/MIME?
- Specification of performance measures for both SMTP and HTTP.
- Review of current "profile" section in the spec with an eye toward
integrating performance considerations into various profiles, e.g.
HTTP>GzipEnvelope>CompB4EncPayloads>EncPayloads
HTTP>GzipAll
SMTP>GzipB64All
...
If I missed anything, please add to this.
-Matt
----------------------------------------------------------------
To subscribe or unsubscribe from this elist use the subscription
manager: <http://lists.oasis-open.org/ob/adm.pl>
