LNPA-WG Slow Horse Subcommittee
9-14-99 meeting at Chicago
[9-21-99 revision]

PARTICIPANTS:

AT&T - H.L.Gowda, Beth Watkins
Ameritech - Donna Navickas
Bell Atlantic - Gary Sacra
BellSouth - Ron Steen, Dave Cochran
Canadian LNP Consortium LLC - Marian Hearn
ESI - Mike Panis, Jim Rooks, Ron Stutheit
GTE - Bob Angevine, Richard Bell, Gary F. Willett
Illuminet - Maggie Lee
Lockheed Martin - Marcel Champagne
MCI WorldCom - Steve Addicks, Gustavo Hannecke, Denise Best
Nextlink - Shelly Shaw,
SBC - Jackie Klare, Charles Ryburn
Sprint - Dave Garner
Stentor - Josee Neron
TSE - John Nakamura
tti - Alan Hancock, Heather Moon
Telcordia - John Malyar
Winstar - Dan Gonos
USWest - Tommy Thompson

COMPLETE PROPOSED LSMS REQUIREMENTS

At last month's meeting, two LSMS requirements were accepted. Only a
short framework of the requirements was developed at that meeting,
however. John Malyar agreed to write up the LSMS requirement
describing NPAC broadcast rates that must be accommodated; Beth
Watkins agreed to prepare the LSMS requirement describing
availability. We planned to finish the text at today's meeting,
including proposing specific values and suggesting how performance
would be measured.

Broadcast Rate Accommodation

The broadcast requirement uses for its sustained rate the published
NPAC capacity as reflected in Exhibit N of the LLC's Master
Contract. The peak rate is expressed as twice the sustained rate,
for a period of up to five minutes, such period occurring not more
than once per hour. The performance described refers to a single
NPAC region. The requirement test of LSMS performance uses the same
figures developed for NPAC capacity determination in Exhibit N: a TN
broadcast mix of TNs so that 20% are sent as individual TNs and 80%
are sent as ranges of 20 TNs each. The proposed text for this
requirement will be distributed in a separate e-mail message next week.
Steve Addicks reported that the Northeast LLC had agreed to ask
Lockheed for data on NPAC broadcast rates. In making its request,
the LLC explained that the ultimate use of this data will be at the
LNPA-WG's Slow Horse subcommittee and asked Lockheed to indicate any
concerns they have about public release of the information provided
in their reply. Lockheed has agreed to respond by September 20th,
the Northeast LLC's next meeting.

Availability

A suggestion was made that downtime that does not actually result in
a broadcast failure should not be included in any availability
requirement. There was not broad support for this approach, however,
apparently because of the complexity of making such a measurement.
There was concern that the best method to measure availability is
unclear. For example, how would measurements be taken in a situation
where the LSMS association is up, but LSMS application is unable to
operate, or in a situation where service provider thinks it's on the
air, but NPAC thinks it's not?

There was suggestion that any LSMS availability standard indicate
the time frame for permissible planned outage as well as the
fraction of downtime permitted. There was interest in allowing
additional downtime to which industry has agreed, such as the
current Sunday morning six-hour maintenance windows during which
service providers have agreed not to port.

The proposed text for this requirement will be distributed in a
separate e-mail message next week.

ROOT CAUSE ANALYSIS

To begin the root cause analysis, we first listed all items which
could cause failure of LSMS to receive an NPAC broadcast:
1. T-1 links - unstable, "bouncing"
2. T-1 links - down
3. LSMS application - down
4. LSMS application - busy (delays exceed 3x5)
5. NPAC application - down
6. NPAC application busy - (general or SP-specific)
7. LSMS CMIP gateway software
8. LSMS router hardware
9. NPAC CMIP gateway software
10. NPAC router hardware
11. SP internal network (firewall, LAN, etc.)
12. NPAC internal network (firewall, LAN, etc.)

The data readily available from the NPAC, such as was presented by
Lockheed in a series of monthly Slow Horse Reports from January
through June 1999, does not lend itself to a root cause analysis.
Consequently the subcommittee asked Lockheed to select and
investigate a sample of broadcast failure events over the next few
weeks and to report their findings October 12th at the next Slow
Horse subcommittee meeting. The Lockheed investigation is to include
contacting the involved service providers if necessary to assure
that the root cause is accurately determined for each broadcast
failure. Marcel Champagne agreed to provide written confirmation of
Lockheed's understanding of the subcommittee's request. A longer
and more thorough investigation will require more formal request,
by an LLC for a Statement of Work.

BEGIN PROPOSED SOA REQUIREMENTS

No work was done at today's meeting on possible SOA requirements.

EXPANDING TEAM'S SCOPE

Steve Addicks, MCI WorldCom asked that the team expand its work to
deal with the end-to-end process, to reflect the business need for
timely update of LNP call routing databases after cutover of a newly
ported end-user, rather than just the portion between the NPAC and
the LSMS. Of those carriers voicing an opinion, only MCI WorldCom
supported attempting to deal with the end-to-end approach now. AT&T,
Ameritech, Bell Atlantic, BellSouth, GTE, Illuminet, Sprint, SBC,
and Winstar felt the subcommittee's work should remain as originally
chartered, to deal the NPAC-LSMS portion of the slow horse problem.
However, SBC and Sprint agreed that eventually the effort could be
expanded to be an end-to-end review. GTE agreed that the SOA-NPAC
aspect also could be included now under the group's charter.

NEXT MEETING

The next meeting of the LNPA-WG's Slow Horse Subcommittee is
scheduled for the morning of 10-12-99, at Kansas City. The two
proposed LSMS requirements should be completed at that meeting and
possible SOA requirements introduced. SOA performance contributions
are invited for consideration prior to next month's meeting.