Ungodly packet loss rates

Mailman List Mirror (Read Only)
1

[Quotes reordered]

If you want good internet service to a particular point on the network,
you must shop for it specifically and be prepared to continue to shop
for it whenever the underlying network topology changes.

My point is that I should be able to expect better than a 40-percent
loss rate to *any* point on the Net, at *any* time, from *any*
provider. In fact, I think I should be able to expect less than 5
percent, and probably less than 1 percent. Back when I actually
touched routers, I used to recommend that people upgrade their
internal networks when loss rates hit 1 percent.

A situation where I have to "shop around" for connectivity depending
on what site I want to talk to that day is just plain unacceptable.

You're transiting two NAPs (MAE-WEST and the DEC exchange (?)) which
both provide service that's essential on one hand (wide access to
the greater Internet) and lousy on the other hand (focussed access
to a particular site).

It doesn't look to me as though the loss is being introduced at the
NAPS. If you look at the trace, you'll see that significant loss
starts to appear within Alternet, well after MAE-west. It looks as
though more loss appears inside BBN's network, although it's difficult
to tell because of the already large Alternet loss.

my suggestion would be to ask cisco what isp they recommend
for dialup access from your location, and dial in through them.
you should be able to lop off 60% of your hops and probably 99%
of your problems if you pick right.

Thanks for the suggestion, but I work for Cisco, so ordinarily I have
a Frame Relay connectivity into Cisco's network. This week the
computer I usually use for work is in for service. The Cisco LAN at my
house is air-gapped from the personal LAN, and it would be a real pain
to reconfigure everything, so for the moment I'm coming in to Cisco
over the Net, using my personal service. It's purely temporary. If
I did interactive work under these conditions on a regular basis, I'd
have gone insane long ago.

                                        -- John B.

2

A situation where I have to "shop around" for connectivity depending
on what site I want to talk to that day is just plain unacceptable.

If you're like most people working off site, there are a few hosts that
you need to keep long TCP sessions open to, and the rest of the net that
gets random traffic.

It doesn't look to me as though the loss is being introduced at the
NAPS. If you look at the trace, you'll see that significant loss
starts to appear within Alternet, well after MAE-west.

Traceroute is less useful a tool than you think in the face of congestive
loss. Routers can and do selectively prioritize the queueing packets
based on their type, and if I were a network operator I would have no
hesitation about dropping traceroute or ping packets to low priority.

If I did interactive work under these conditions on a regular basis, I'd
have gone insane long ago.

No, you would have changed ISPs a long time ago, instead of whining to
NANOG that the entire world is providing sub-standard service.

Ed

3

My point is that I should be able to expect better than a 40-percent
loss rate to *any* point on the Net, at *any* time, from *any*
provider. In fact, I think I should be able to expect less than 5
percent, and probably less than 1 percent.

Honestly, here I think you're smoking something. Whatever it is, it
must be REAL good.

(And from a coworker, a fellow sysadmin:
  "There is something to be said for the idiot; if the world was
   void of idiots, how then would *we* look?"
  How in the hell can you expect a 100% success rate over (1) a slow
  modem link, and (2) to *ANY* site on the world. Hell, do you have
  any *CLUE*--I know you don't--how many sites on the net have servers
  behind 28.8 links??? How great a packet loss do you expect when you
  access them?? Is that provider dependent??? *ANY* site--really?
- mikedoug@texas.net -
   Michael Douglass )

Back when I actually
touched routers, I used to recommend that people upgrade their
internal networks when loss rates hit 1 percent.

Do you have a clue? Due to the structure of the Internet, you
are *going* to have loss. It's inherent in the design of the
system, and unavoidable. And for people upgrading their internal
networks when packet loss hit 1%, woah, sounds like you must
work in Sales.

A situation where I have to "shop around" for connectivity depending
on what site I want to talk to that day is just plain unacceptable.

No, I beleive the person who recommended that suggested you shop around
for the best provider *to start out with*, not bitch, whine, and moan
when your connection is not 100% perfect through the one you
currently have.

It doesn't look to me as though the loss is being introduced at the
NAPS. If you look at the trace, you'll see that significant loss
starts to appear within Alternet, well after MAE-west. It looks as
though more loss appears inside BBN's network, although it's difficult
to tell because of the already large Alternet loss.

Traceroute is *not* a good tool to diagnose packet loss problems.
I've had traceroute tell me that a packet loss problem was between
two points 3-4 hops "out", when actually it was with the T-1 at
my site, the "first hop" in the trace.

Thanks for the suggestion, but I work for Cisco, so ordinarily I have
a Frame Relay connectivity into Cisco's network. This week the
computer I usually use for work is in for service. The Cisco LAN at my
house is air-gapped from the personal LAN, and it would be a real pain
to reconfigure everything, so for the moment I'm coming in to Cisco
over the Net, using my personal service. It's purely temporary. If
I did interactive work under these conditions on a regular basis, I'd
have gone insane long ago.

I dont see where a temporary network problem such as you describe should
result in a message being sent to the various ISPs and the NANOG list.
My suggestion: quit bitching and wait for your FR connection to be
restored, or reconfigure your current equipment (if you work at Cisco,
it shouldn't be TOO hard).