SI
SI
discoversearch

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor.  We ask that you disable ad blocking while on Silicon Investor in the best interests of our community.  If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.
Technology Stocks : Newbridge Networks -- Ignore unavailable to you. Want to Upgrade?


To: pat mudge who wrote (12732)8/15/1999 1:00:00 AM
From: Tunica Albuginea  Read Replies (1) | Respond to of 18016
 
Pat:WCOM/LU problems:?is answer in this fumble/Peppe 6/8/98 conversation I just unearthed?
Perhaps the key is at the end:

Perhaps, over time, CSCO, LU and NT will come up with something.
Since the telecom market is standards based,
it won't happen soon.
At the moment though they are relying on a confused marketplace and are stepping up their PR volume to protect
their niche.


TA

============================================================
To: fumble who wrote (5023)
From: fumble (5024)http://www.techstocks.com/stocktalk/readmsg.aspx?msgid=4771951
Monday, Jun 8 1998 3:01PM ET
A piece of NYTimes article:
nytimes.com

<<"There's no one silver bullet," said Bill O'Shea, president of Lucent's data networking business. "It's years and
years of experience, building systems and watching how they operate" that insure reliability and avoid failures.


Yet Lucent and Nortel are still very much newcomers. Only this month will Lucent begin testing its first
billion-bits-a-second packet switch based on Internet protocols,
which could become a true competitor to Cisco's
most advanced gear. Northern Telecom, for its part, derived only $785 million of its $15.4 billion in revenue last
year from data networking; it has a long way to go to be on an equal footing with Cisco. >>

I thought that NN's 36170 is rated at 12.8 Gbits/sec, and slated to go to 50Gb before the year is ended. 36190 is
much faster than that.


Why even mention LU's 1 Gb switch. Is my arithmetic off? Apples and Oranges?
===========
To: fumble (5024)
From: Peppe, Jun 8 1998 3:35PM ET
Reply # 5027 techstocks.com

<Why even mention LU's 1 Gb switch. Is my arithmetic off? Apples and Oranges?>

Yup. LU and CSCO's switches are layer 3 devices, also known as routers. NN's 36170 and 36190 are ATM
layer 2 devices, with far less intelligence, therefore can be made to be faster. (Less overhead,etc.)

=============
To: Peppe
From: fumble ( 5046)http://www.techstocks.com/stocktalk/readmsg.aspx?msgid=4785407
I thought the ATM speed came because of smaller packets which could be handled with silicon rather than
software. Intelligence? Doesn't LU and CSCO's devices just throw away packets when busy, whereas ATM
provides hooks to ensure that Quality of Service and bandwidth are contractually guaranteed.
This would imply a
bit more 'usable' intelligence with ATM.
=============
To: fumble who wrote (5046)
From: Peppe, Jun 9 1998 1:55PM ET
Reply # 5048 techstocks.com
Fumble,

Routers are IP layer 3 devices that hold information of where a packet should be directed. An ATM switch
is a cell based layer 2 device that passes cells to its next hop.
Basic 'routing" is being added to layer 2 switches
via MPOA or MPLS. However, this is to route amongst ATM switches and does not scale the same as say OSPF or
BGP4,
the internet's routing protocols. Therefore, the "intelligence" resides on the edge, within the routers.

Routers, being software based today, are much slower than ATM switches or layer 3 switches, which can both
perform at 'wire speed',
depending on who you believe (vbg).

QoS is superior today in ATM because you can actually carve out a portion of your connection and dedicate it to a
customer or application.
QoS on routers is developping, but not there yet. ATM QoS schemes tend to be
proprietary, however.

I wasn't slamming ATM, just trying to answer your question.

====================
To: Peppe who wrote (5048)
From: fumble Jun 9 1998 3:51PM ET
Reply # 5050 techstocks.com

Peppe

ATM and IP networks are different, so the description of 'intelligence' needs to be different too.

Router intelligence is basically a big table lookup.
Each packet coming through an IP router is a 'new' packet. No memory of previous packets for the same customer
is recorded. The router looks at the destination address, checks in its table, finds the port leading on to the
destination, and sends the packet on its way. As networks become more complex, the routing tables get longer and
longer, thus scalability is in question for IP networks.


BGP4 and OSPF are ways that each router contacts its neighbor routers to update any routes which may have
gone bad since the last update. These IGPs and EGPs are not used to route individual user packets.

ATM on the other hand, is a connection oriented transport protocol. The first few packets coming for a customer
contain destination routing information. These packets 'set up a connection' for following packets. This connection
uses a quicky routing scheme which does not require the ATM switches along the route to dig through long routing
tables, only shorter tables for each of its ports and each of the virtual connections it is handling at the moment. The
last few cells of a connection contain information which allows the connection to be torn down and the memory and
table space to be used by new connections.

Since ATM connection setup packets contain information about the QoS requirements and the bandwidth
requirements for that connection, those aspects can be supported.

Telephone connection packets slide onto the ATM networks with appropriate specs so that telephone connections
are clear and do not contain sporadic pops, clicks, and pauses.


IP data packets move through a MPOA (or other) stage at the edge of the ATM network where appropriate
ATM specs are added to create a virtual connection for those packets. Since data packets usually do not require a
timely delivery, IP traffic can be bumped to allow cells with higher QoS specs (and higher revenue) through the
ATM network.

IP packets containing telephone conversations are problematic.

As telcos learn how to control their IP traffic better, I would imagine that IP packets containing telephone
conversations will become more erratic (data packets do not have to be delivered at a particular rate..),
thus that
area of telco competition will be relegated to the 'back of the bus', or required to pay a bit more for timely delivery
along contracted virtual pathways. ATM is the future.

Regards

fumble
======================
To: fumble who wrote (5050)
From: Peppe Jun 9 1998 4:00PM ET
Reply # 5052 techstocks.com
Fumble,

Obviously you know something about ATM. I won't debate who will "win" in the core, because I don't care. I
agree with you that IP voice is problematic...today. I'm betting that CSCO, LU and NT figure out a way to solve
that issue, over time. They all have heavy investments in high speed IP routers and all have ATM products in the
mean-time.


Cheers,

Peppe

========================
To: Peppe who wrote (5052)
From: fumble
Jun 9 1998 4:41PM ET
Reply # 5055 techstocks.com
Peppe,

Perhaps, over time, CSCO, LU and NT will come up with something.

Since the telcom market is standards based,
it won't happen soon.

At the moment though they are relying on a confused marketplace and are stepping up their PR volume to protect
their niche.


As you may have guessed, I'm long on NN.

Regards,

fumble
============
Peppe, reply 5057 techstocks.com

============================================================

eom

TA

============================================================

you said

More WCOM coverage, this with reference to the CBOT issue becoming a legal matter:

computerworld.com

And from the New York Times:

>>>>
August 14, 1999

Network Trouble Enters Eighth Day
By REUTERS
EW YORK - Long distance phone company MCI WorldCom Inc.'s high-speed data network continued to suffer
problems Friday, idling the Chicago Board of Trade's (CBOT) electronic trading system and disrupting service to
thousands of businesses for more than a week.

MCI WorldCom's network problem began eight days ago due to a glitch in software provided by Lucent
Technologies Inc.

--------------------------------------------------------------------------------
Related Article
MCI Network Problems Persist
(August 13, 1999)
MCI Still Fixing Glitches
(August 12, 1999)

MCI Worldcom Problems Hit Its Data Network Users
(August 10, 1999)

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

The outages affected about 15 percent of MCI WorldCom's network and an estimated 70,000 customers such as
the CBOT, Internet service providers, banks and other businesses that rely on high-speed frame relay networks
to exchange large amounts of computer information in short and frequent bursts.

The CBOT, the largest U.S. derivatives exchange, said Friday its electronic trading system, Project A, would
remain shut until Sunday night when the exchange moves to its backup system indefinitely.

"Market users worldwide depend on Project A around the clock and MCI WorldCom has let them down for one
full week. As a result of MCI WorldCom's failure to deliver on their promises to me early last week, the CBOT is
pursuing all available remedies," CBOT President Tom Donovan said in a statement.

Another customer, a major U.S. retailer, said MCI WorldCom's network service began to come back during the
middle of this week only to grind to a halt again Friday morning.

"Things started to get better, but there was some failure overnight and now we are in worse shape than ever,"
said one MCI WorldCom customer who declined to be named.

The customer said the network problems prevented him from communicating with distributors, other company
locations and his clients.

"I've never seen a network outage last a week. If they had been more clear in the beginning about the extent of
the problem or their ability to handle it, we could have made other contingency plans. But now we're just going
with day-by-day information and that's not helping," the customer said.

Clinton, Miss.-based MCI WorldCom said the network is now stable, but problems arose last night during some
network maintenance. That instability may have made service problems worse for some customers, but service
quality should improve today, said MCI WorldCom spokeswoman Linda Laughlin.

The company said it continues to restore service to customers and to talk to clients one-on-one about the
problems and any potential compensation.

MCI WorldCom's outage followed a similar problem at AT&T Corp. last year. AT&T and other companies such as
Qwest Communications International Inc. said they have gained customers this week as some disgruntled MCI
WorldCom clients searched for new service providers.

>>>>>