vixie@mfnx.net said:
I just liked the description, "small tier-1".
(I guess if you have a PC and a modem, you're tier-1 now?)
If you refuse to peer with anyone at all, you can be tier-0. This
can be achieved with considerable savings to phone line utilization.
Actually, we already have a tier-0. See:
http://www.opnix.net/perl/PressRelease.cgi?article=100032
(And many other things on their website.)
Particularly amusing is:
http://www.opnix.net/whatwedo/performance.shtml
--msa
Particularly amusing is:
If you have any questions about our route intelligence technologies (the
above link talks about that a bit) or how it relates-to /
interoperates-with BGP4, please feel free to ask me. I'd be happy to
answer any questions. 
InterNAP has done the tier-0 marketing dance for some time. Quite
successfully, as a matter of fact. Secret Sauce sells like hotcakes. Wall
Street likes it as well. Not much of a performance increase, though.
- Daniel Golding
Im particularily intrigued by the Opnix customer care practices. A
late-nite call to the Opnix NOClet on duty to find out why a certain
netblock was unable to connect to the Opnix webserver(s) received the
response "Are you a customer? I dont recognize your voice..." How does
this scale beyond 10 customers?
On a semi-technical front, what are you using to monitor RTT and how do
you switch to new paths when you see that a current path is
congested/latent/etc? How do you adapt to changes without thrashing, and
how do you handle multi-homed customers, particularily those that are
multihomed to multiple Opnix PNAPS^H^H^H^(oops, wrong marketing hype
engine enabled)POPS?
-troy
late-nite call to the Opnix NOClet on duty to find out why a certain
netblock was unable to connect to the Opnix webserver(s) received the
response "Are you a customer? I dont recognize your voice..." How does
I've never heard of that one. Likewise, that's a crappy response. I'll
forward this to our NOC Manager and CTO. Nobody's perfect, but the reply
you got is totally unacceptable. I can assure you we don't handle calls
based on voice recognition.
On a semi-technical front, what are you using to monitor RTT and how do
All of our tools/systems/protocols/etc... are proprietary, but just from
the RTT perspective, traceroute or ping can give you a very basic RTT
measurement.
Speaking of tools, we're releasing an open source utility called "OpRoute"
at the end of this month. The tool works something like traceroute, but it
also reports # of NAPs, AS hops, layer 3 hops, latency, packet loss,
etc... It also gives you the option to show a side-by-side compariason of
those statistics (from "a" to "z") on your network versus another outside
network.
Anyway, this is coming out (with source code) at the end of the month --
if interested, I'll post a notice to this list when its released.
you switch to new paths when you see that a current path is
congested/latent/etc? How do you adapt to changes without thrashing, and
how do you handle multi-homed customers, particularily those that are
"Thrashing" routes is a concern. We have to throttle-back our
optimizations and have put in some algorithms for dealing with this. The
actual process of changing the route we do with proprietary protocols
in-house.
As for multihomed customers, that gets me into a whole new area... I don't
know if this list would care to hear the entire bit, so I'll just hit the
basics. If anyone wants more information, just email me off-list. Anyway,
our current product (bandwidth) is being augmented by a new product
(called IRIS) that goes into beta in April. IRIS is basically a
client-side version of the routing intelligence technologies which
interoperates with our core. IRIS is specifically for diverse networks
(multihomed or otherwise) -- no matter if they're using the Opnix
bandwidth product or not.
multihomed to multiple Opnix PNAPS^H^H^H^(oops, wrong marketing hype
engine enabled)POPS?
Heh, you can never keep the marketing department under control.
~Jay
Hmmm.
How do you figure out when you have crossed a NAP? Hard-coded table of
exchange point IP blocks? That will work for the larger NAPs, but wouldn't
necesarily detect passing over a peering switch in a PAIX or Equinix
facility (unless they are all in your table...) Munging reverse DNS?
Of course, treating all NAPs the same is tricky business. FDDI, ATM, Gig
Ethernet, etc, are far different animals. Most of the conventional wisdom
surrounding public peering came to light during the heyday of the FDDI
NAPs.
Daniel Golding NetRail,Inc.
"Better to light a candle than to curse the darkness"
I believe the "NAP Detection" algorithms are proprietary (I'll have to
check with our VP of Software Devel to be sure). I do know for sure,
however, that we do not just use a static table.
Regardless of the architecture of a NAP (or any of the connections
therein), the act of traffic simply going through a NAP does not always
mean that's a bad route. Many (most) of the NAPs are overloaded, sure, but
what really matters is latency, packet loss, etc... If a NAP route has the
highest performance (hey, it's possible then it's a good route. Many
times this may not be the case, but it's possible.
~Jay
"BGP4 makes static routing decisions"
The last time I checked, BGP4 didn't do _ANTHING_ static. Perhaps the
kind folks at OPNIX would like to tell us something we don't know about
BGP. After all, they did manage to create a new tier.
Yes Jay. I am interested in your technology. Beyond the fact that it
appears to be 100% marketing hype BS, can you elaborate, _WITHOUT NDA_
about the following:
(1) How does your "Best Path Global Transit (tm)" interact with the GLOBAL
STANDARD BGP4 with respect to:
(a) Influencing INBOUND data
(b) Influencing INBOUND data
(c) Influencing INBOUND data
(d) Influencing INBOUND data
(e) Influencing INBOUND data
(f) Influencing INBOUND data
(g) Influencing INBOUND data
(h) Influencing INBOUND data
(i) Influencing INBOUND data
(j) Influencing INBOUND data
(k) Influencing INBOUND data
(l) Influencing INBOUND data
(m) Influencing INBOUND data
(n) Influencing INBOUND data
(o) Influencing INBOUND data
(p) Influencing INBOUND data
(q) Influencing INBOUND data
(r) Influencing INBOUND data
(s) Influencing INBOUND data
(t) Influencing INBOUND data
(u) Influencing INBOUND data
(v) Influencing INBOUND data
(w) Influencing INBOUND data
(x) Influencing INBOUND data
(y) Influencing INBOUND data
(z) Influencing INBOUND data
(2)Is your "Best Path Global Transit (tm)" going to be released to public
domain or are you going to become a "Tier-0" Island.
(3) Have you thought about how much the rest of the world is going to
laugh at "Tier-0?"
ROTFLMAO
Yes Jay. I am interested in your technology. Beyond the fact that it
appears to be 100% marketing hype BS, can you elaborate, _WITHOUT NDA_
about the following:
A corporate web site, most often, is meant to be a marketing tool. We all
know that marketing and engineering doesn't always mix well. Rather than
dwelling on the flame-bait, I'll just address your questions as best I
can. Some of the information is proprietary, and will not be disclosed
with, or without, an NDA.
(1) How does your "Best Path Global Transit (tm)" interact with the GLOBAL
STANDARD BGP4 with respect to:(a) Influencing INBOUND data
[snip]
(z) Influencing INBOUND data
We influence (NOT control) inbound data a number of ways (some
proprietary, some not). One of which is that we purchase all of our
transit from the major carriers. With this relationship, we can get them
to "do the right thing" on their end(s).
(2)Is your "Best Path Global Transit (tm)" going to be released to public
At this time, Best Path Global Transit will continue to be held as a
proprietary technology. It does fully interoperate with BGP, but will
remain proprietary for now.
(3) Have you thought about how much the rest of the world is going to
laugh at "Tier-0?"
I have already addressed this point in a later reply on this thread, but
I'll say it again... I don't agree with the "Tier-n" stuff, regardless of
what the "n" is. Our web site is currently in the process of a re-do, and
I'm certain that terminology will not be in the new version. Sometimes
trying to contain marketing == herding cats.
~Jay
yOn Mon, 19 Feb 2001, Jay wrote:
Speaking of tools, we're releasing an open source utility called "OpRoute"
at the end of this month. The tool works something like traceroute, but it
also reports # of NAPs, AS hops, layer 3 hops, latency, packet loss,
etc... It also gives you the option to show a side-by-side compariason of
those statistics (from "a" to "z") on your network versus another outside
network.
And just how do you know when someone went through a NAP?
Anyway, this is coming out (with source code) at the end of the month --
if interested, I'll post a notice to this list when its released.
Very interested. I _REALLY_ get off on screwing with marketing hype
motivated people.
"Thrashing" routes is a concern. We have to throttle-back our
optimizations and have put in some algorithms for dealing with this. The
actual process of changing the route we do with proprietary protocols
in-house.
Read:
Modify route-map xxx
"clear ip bgp x.x.x.x soft in"
As for multihomed customers, that gets me into a whole new area... I don't
know if this list would care to hear the entire bit, so I'll just hit the
basics. If anyone wants more information, just email me off-list. Anyway,
our current product (bandwidth) is being augmented by a new product
(called IRIS) that goes into beta in April. IRIS is basically a
client-side version of the routing intelligence technologies which
interoperates with our core. IRIS is specifically for diverse networks
(multihomed or otherwise) -- no matter if they're using the Opnix
bandwidth product or not.
Let me guess. IRIS doesn't run on the 75xx or 120xx?
Rules of thumb in internet marketing to geeks:
(1) Make it work with Cisco
(2) Make it ACT like Cisco
> multihomed to multiple Opnix PNAPS^H^H^H^(oops, wrong marketing hype
> engine enabled)POPS?Heh, you can never keep the marketing department under control.
~Jay
It's real easy when you have ENGINEER types holding the purse. You don't
let the Marketing Types release BULLSHIT to the word. If/When they do,
you fire their dumb %$$es. It's that simple. They have released
something that is going to be an EMBARRASSMENT to the company in front of
anyone who is technically compitent. They have alienated 90% of your
customer base. They are FIRED.
The last time I checked, BGP4 didn't do _ANTHING_ static. Perhaps the
kind folks at OPNIX would like to tell us something we don't know about
BGP. After all, they did manage to create a new tier.
"Static" certainly isn't the right word -- I'll agree with that. The
marketing department apparently liked it. Again, just as with all the
"Tier-n" garbage, that language will be changed in the new version of the
web site.
However, the point is simply that BGP4 makes routing decisions based on AS
hops. Yet, AS hops really don't mean anything when it comes to
performance. The stuff that matters is things like latency, packet loss,
etc... Granted, more AS hops MIGHT have more latency/packet loss, but
that's not always the case.
> Yes Jay. I am interested in your technology. Beyond the fact that it
> appears to be 100% marketing hype BS, can you elaborate, _WITHOUT NDA_
> about the following:A corporate web site, most often, is meant to be a marketing tool. We all
know that marketing and engineering doesn't always mix well. Rather than
dwelling on the flame-bait, I'll just address your questions as best I
can. Some of the information is proprietary, and will not be disclosed
with, or without, an NDA.
Sweet.
Read: "Our marketing department came up with this. Engineering is still
trying to figure out what the ^&*K they are talking about. As a result,
we are not at liberty to discuss it, period, the end."
> (1) How does your "Best Path Global Transit (tm)" interact with the GLOBAL
> STANDARD BGP4 with respect to:
>
> (a) Influencing INBOUND data
[snip]
> (z) Influencing INBOUND dataWe influence (NOT control) inbound data a number of ways (some
proprietary, some not). One of which is that we purchase all of our
transit from the major carriers. With this relationship, we can get them
to "do the right thing" on their end(s).
So, what you're saying is that you have come up with a _better_
SAVVIS/INTERNAP model. OK. Why don't you just have your marketing
monkeys say that?
> (2)Is your "Best Path Global Transit (tm)" going to be released to public
At this time, Best Path Global Transit will continue to be held as a
proprietary technology. It does fully interoperate with BGP, but will
remain proprietary for now.
I know. I know. It's hard to release a "vapor" protocol. You need
Microsoft Marketing to do that.
> (3) Have you thought about how much the rest of the world is going to
> laugh at "Tier-0?"I have already addressed this point in a later reply on this thread, but
I'll say it again... I don't agree with the "Tier-n" stuff, regardless of
what the "n" is. Our web site is currently in the process of a re-do, and
I'm certain that terminology will not be in the new version. Sometimes
trying to contain marketing == herding cats.~Jay
I'll give you credit for this much. You have someone subscribed to NANOG
and someone up when the rest of us who REALLY run networks have the time
to have an email conversation.
_AND_ You have made me laugh after a REALLY long and stressful day.
And just how do you know when someone went through a NAP?
...read the code when OpRoute is released at the end of this month.
Very interested. I _REALLY_ get off on screwing with marketing hype
motivated people.
It would seem you _REALLY_ get off on screwing with anyone that does not
think the same way you do. If you would like to carry-on a flame war with
me, why don't we take it off list.
Let me guess. IRIS doesn't run on the 75xx or 120xx?
No, it doesn't. It does, however, fully integrate with BGP4-enabled
routers -- Cisco, Juniper, Lucent, Nortel, Foundry, Riverstone, etc...
Rules of thumb in internet marketing to geeks:
(1) Make it work with Cisco
(2) Make it ACT like Cisco
Thanks for the pointers; already covered though. IRIS does work just fine
with Cisco gear and, just like most other network gear CLIs, the IRIS CLI
should act similar to IOS.
It's real easy when you have ENGINEER types holding the purse. You don't
Four of the five founders of Opnix (including myself) are "engineer
types."
let the Marketing Types release BULLSHIT to the word. If/When they do,
you fire their dumb %$$es. It's that simple. They have released
It seems someone with a marketing degree must have pissed in your
Cheerio's or something. Like any other group of people, there are both
good and bad within the group. Like I said, engineering and marketing
usually don't mix well. We all know that. We get plenty of opportunities
to flame marketing and sales people for SPAM on lists -- we probably don't
need to put any more effort into it.
Again, if you want to have a flame-war, let's take it off-list.
~Jay
It would seem you _REALLY_ get off on screwing with anyone that does not
think the same way you do. If you would like to carry-on a flame war with
me, why don't we take it off list.
Not worth the time. You're not a marketing type. Just an
engineer/investor who got screwed by giving marketing too much latitude.
> Let me guess. IRIS doesn't run on the 75xx or 120xx?
No, it doesn't. It does, however, fully integrate with BGP4-enabled
routers -- Cisco, Juniper, Lucent, Nortel, Foundry, Riverstone, etc...
So, nothing new eh?
> It's real easy when you have ENGINEER types holding the purse. You don't
Four of the five founders of Opnix (including myself) are "engineer
types."
You obviously didn't have your marketing monkeys pass your marketing
hype through the "reality filter." Had you done so, we wouldn't be having
this discussion.
> let the Marketing Types release BULLSHIT to the word. If/When they do,
> you fire their dumb %$$es. It's that simple. They have releasedIt seems someone with a marketing degree must have pissed in your
Cheerio's or something. Like any other group of people, there are both
good and bad within the group. Like I said, engineering and marketing
usually don't mix well. We all know that. We get plenty of opportunities
to flame marketing and sales people for SPAM on lists -- we probably don't
need to put any more effort into it.Again, if you want to have a flame-war, let's take it off-list.
~Jay
Jay, It is very amusing that you take this personally and wish to take
things off-list. IMHO, if your product is truely as revolutionary to
networking as your marketing monkeys would have the "know nots" believe,
any discussion of it will be truely on-topic for NANOG and in will in-fact
become a MUST-READ for anyone who wishes to survive in the new networking
world that your company has created.
If you have nothing more than marketing hype to give us, shut up. This is
an informational list. We don't give a rats behind about pipe-dreams and
"we wish it were a standard" protocols. We run _REAL_ networks. Nets run
on "proprietary" protocols are _BY_DEFINITION_ islands and as such, NOT
part of the global internet.
Trademark, copyright, have your dog leave his scent on the source/roadmap
of your "proprietary protocol" but until you release it to the GP so we
can critique it, you're nothing more than a rainmaker to me.
Fair enough. I refuse to get into personal disputes on-list. However, your
points in this recent thread are well-taken. I need to get some sleep now
(have to be up in ~3 hours), but I'll certainly reply to this post
tomorrow.
~Jay
4 atm8-0-093.CR-2.uschcg.savvis.net (64.241.88.65) [AS 6347] 16 msec 12 msec 16 msec
5 at-0-2-0902.uslsan2-j20c.savvis.net (64.242.22.134) [AS 6347] 56 msec 60 msec 56 msec
6 opnix-1.usphnx.savvis.net (64.241.66.130) [AS 6347] 76 msec 80 msec 76 msec
7 bi01-ve2.phx.opnix.net (216.183.192.3) [AS 18591] 76 msec 76 msec 76 msec
8 www.opnix.net (216.183.194.131) [AS 18591] 80 msec 76 msec 76 msec
I'd be really torqued off if I were you. It would appear that you're a
direct customer of SAVVIS and that 76 ms RTT time to your network from
another SAVVIS peer is outside of tha SAVVIS SOA.
Not that it's any better from an accepted Tier_1:
10 opnix-dw.customer.alter.net (157.130.245.2) [AS 701] 76 msec 76 msec 76 msec
11 bi01-ve2.phx.opnix.net (216.183.192.3) [AS 18591] 76 msec 72 msec 72 msec
12 www.opnix.net (216.183.194.131) [AS 18591] 72 msec 76 msec 76 msec
Globalcenter, UUNET, Level 3, C&W, and SAVVIS? That's all you've got?
I'll put my connection at home up against that!
I'm guessing that your "Best Path Global Transit (tm)" was on the
fritz.... Right?
Opnix, Inc. (ASN-OPNIX)
2220 W. 14th St.
Tempe, AZ 85281
US
Autonomous System Name: OPNIX
Autonomous System Number: 18591
Coordinator:
Kloain, John (JK928-ARIN) john@opnix.com
480-966-7551
Record last updated on 15-Sep-2000.
Database last updated on 19-Feb-2001 18:27:46 EDT.
Whoa.... High number ASN. MUST be a revolutionary,
[snip]
(3) Have you thought about how much the rest of the world is going to
laugh at "Tier-0?"
I could spend days pointing out some "new" companies and their
rather funny marketing stuff. Personally, I think some of the
marketing "brandings" which are out there now border on possible
false advertising / false representation, but I digress..
Adrian
Jay,
You'll be hard pressed to find a border router of a BGP speaker that makes
routing decisions based soley on AS path length. There are many variables
that a peer can take into account to decide best-path and most of us use
those variables to "shape" traffic in some way-shape-form.
I suspect that your new technology simply adds another variable to the
best-path-selection process based on your ping/traceroute script RTT's to
a specific network based on what you have divolged thusfar.
If this is in fact the case, I have two problems with the technology:
(1) To interwork with existing routers, it has to do ebgp-multihop to
peers and then send its tables to the "REAL routers." This adds a point
of failure to the mix. (subtract several 9's)
(2) In order to have an ACCURATE picture of the global internet, your
software will have to constantly probe remote networks to establish the
best RTT path which in turn will create N^routing_table_size traffic on at
least ONE link (if you do it the lazy way) and N*X^routing_table_size
traffic if your do it from multiple points to establish a "mean RTT".
Both of the above have CONS which outweigh any potential benefit that may
be obtained and the second has the potential, if deployed on any large
scale, to create a 1:10,000(+) S:N ratio on the internet as a whole as
every router aggressively seeks out a better path to the each network.
My alcohol induced deductions based on the limited information you have
provided lead me to believe that your product, if it in any way is like I
imagine it to be, will do more harm than good and as such, should be
squashed like a fly.
Feel free to embarrass me by releasing the protocol and showing that what
you're doing is in no way like what I have outlined above.