Re: [lmap] LMAP framework issue #1 User-initiated Measurement Tasks

["MORTON, ALFRED C (AL)" <[e-mail]>]

In what we now call case 2, the user-deployed measurement system:

Phil wrote:
... There are at least two ways user-initiation could happen, in outline
<snip case 1>
* a user could deploy their own measurement system, with their own MA, Controller and Collector
(possibly with all three functions in the same box). The user may also want to report
Measurement Results to a third party. One possible situation is that the home contains
a user-controlled MA and an ISP-controlled MA; then the test traffic of one MA is
treated by the other MA just like any other user traffic.
Note that a single MA is instructed by a single Controller and is only in one measurement system.

The issue I see is that an LMAP system of Controller, MA and Collector is incomplete.
In the active measurement scenario, the user's MA needs a remote partner in most cases.
Sure, you could run the DNS measurement, or GET a web page. But to measure anything interesting
the user needs a partner: a Measurement Peer who runs a compatible test protocol
(unspecified by LMAP).

-  In user-deployed case 2, the Measurement Peer (MP) is likely to be operated by a
   different organization (users have limited span of control)

-  The Measurement Peer's org. is likely to have access to some test results
   (through timestamps) and certainly an IP address. Users give-up some info
   to get access to a Measurement Peer.

So, case 2 seems to stretch the bounds of LMAP, because the LMAP-unspecified measurement protocol
between MA and MP spans org. boundaries, and conveys measurement info and user info.
Today's web-based testing services provide a partner IP addrs and a let you see the results,
but they keep copies of those things too.

How can we negotiate the charter's privacy requirements in user-deployed case 2?
We might have to accept that the measurement protocol gives some info away.
I'm interested to learn more from folks with experience in this area.

Al


From: [e-mail] [mailto:[e-mail]] On Behalf Of [e-mail]
Sent: Tuesday, September 03, 2013 12:59 PM
To: [e-mail]
Subject: [lmap] LMAP framework issue #1 User-initiated Measurement Tasks

We've now started creating an LMAP framework doc that merges 3 I-Ds (terminology and the 2 framework docs) - hoping it could be the basis for a WG doc - as mentioned in Berlin.

One section will be about proposed constraints /assumptions - extending http://tools.ietf.org/html/draft-eardley-lmap-framework-02#section-3

I'm going to send a series of emails to try and capture where I think the discussion got to in Berlin &/or propose text for the I-D &/or generate discussion on open issues.

Constraint: User-initiated Measurement Tasks out of scope of LMAP WG
We expect LMAP & IPPM functionality to be used for user-initiated Measurement Tasks, but the WG will not define how. There are at least two ways user-initiation could happen, in outline
* a user could somehow (perhaps via a webpage) request the ISP- (or regulator-) run measurement system to test his/her line. The ISP (or regulator) Controller would then send an Instruction to the MA in the usual LMAP way. The Measurement Results could be returned back via the webpage. Note that a user can't directly initiate a Measurement Task on an ISP- (or regulator-) controlled MA in their home
* a user could deploy their own measurement system, with their own MA, Controller and Collector (possibly with all three functions in the same box). The user may also want to report Measurement Results to a third party. One possible situation is that the home contains a user-controlled MA and an ISP-controlled MA; then the test traffic of one MA is treated by the other MA just like any other user traffic. Note that a single MA is instructed by a single Controller and is only in one measurement system.
For further details see http://www.ietf.org/mail-archive/web/lmap/current/msg00714.html and related messages.

Comments?
Thanks
phil