x
Optical/IP

A Double Dose of IP QOS

Light Reading is serving up a double dose of information on IP QOS (quality of service) today, with the publication of a report covering the basic issues facing service providers and with a free Webinar (online presentation) reviewing the latest developments in silicon chips supporting QOS in telecom equipment.

The report -- IP Quality of Service -- says that the argument over whether QOS is needed in IP networks has subsided for the moment. The folk promoting the "big pipe" idea of simply throwing bandwidth at the whole issue of guaranteed performance have gone quiet (although they haven't gone away), and the real debate now centers on how and where to implement QOS.

Having said that, the report goes on to demonstrate one of the main arguments against QOS -- that it can be fiendishly complicated, and that this can undermine the economic case for employing it in the first place. It also notes that the whole idea of QOS -- enabling telecom operators to offer different grades of service at different prices -- only really works on closed networks right now. Efforts to extend performance guarantees over multiple networks are in their early stages.

All the same, plenty of operators are looking to QOS as a way for them to start generating profits from their IP infrastructure. The report, authored by Graham Finnie, an independent consultant, steps through the technologies that promise to make this possible, outlining recent developments.

Today's Webinar takes things several steps further, by drilling down into the details of how QOS is implemented in telecom equipment using so-called traffic manager chips, a type of network processor. Latest developments in this field are reviewed.

The main presenter in today's Webinar is Simon Stanley, founder and principal analyst at Earlswood Marketing Ltd., a U.K. consultancy, and presenter of several other Light Reading Webinars on different types of communications chips and subsystems. The other speakers come from the event sponsors -- Azanda Network Devices and Motorola Inc. (NYSE: MOT).

There's still time to register for today's Webinar, which will be staged at 2PM Eastern time, 11AM Pacific time. Click here to register, and remember to tune in early to avoid the rush.

— Peter Heywood, Founding Editor, Light Reading
www.lightreading.com Want to know more? The big cheeses of the optical networking industry will be discussing QOS at Lightspeed Europe. Check it out at Lightspeed Europe 02.

DoTheMath 12/4/2012 | 9:34:53 PM
re: A Double Dose of IP QOS
The article alludes to the "fiendish complexity" of QoS
making the economics doubtful. That is the *core* point.

In most mature industries Class of Service or
Quality of Service distinctions go away and give
rise to a "good enough" level of service,
very attractively priced.

Southwest and JetBlue are doing it in the airline
industry, Wal-Mart/Costco in the retail industry (compare them
to Nordstrom and Macys), Dell in the PC industry
(compare commodity priced PCs to value-based mainframe,
unix systems), and so on.

Think about it: even Bill Gates gets the same phone
service (wireline or wireless) and goes to the same
McDonalds that the average American sees. He just
doesn't have the option of paying a lot more to get
better service - I don't know if such a thing exists.

So I consider the entire QoS argument as economically
dead-on-arrival, regardless of the technical interest.
It is an "interesting" engineering problem, giving rise
to tons of PhDs, but I believe the economics dictates
uniform, good-enough service at a price the masses can
afford.

Numerous other industries have shown the way, and I
believe data networks won't be any different. Operators
will not be able to charge the premium to justify all
the investments in QoS; even corporate customers will
flock to cheap commodity services, undercutting the
economic basis of investment in QoS.
mrcasual 12/4/2012 | 9:34:51 PM
re: A Double Dose of IP QOS While I agree that most of the QoS propellerheads out
there don't get the fact that "good enough" is in
fact actually good enough there are certain levels
of service or applications which require better QoS
than what exists today.

Don't forget that QoS can encompass lots of different
measures such as latency, jitter, data loss, and
throughput.

Most people find the QoS of their cellphones acceptable
for occaisonal or convenience calls. Most people also
find the QoS of cell phones unacceptable for their
primary business phone to expand on your Bill Gates
analogy.

The same is true for data services (or voice services
over the data network for that matter. Some people
can live with 56K dial up. Others need a dedicated T3
to connect their remote sites.

The benefits of QoS can be that a single network
infrastructure (i.e. the Internet) can service many
diverse needs if some form of end to end (and end
to end is a requirement) QoS exists. Without it we
end up with the hodge podge we have today.
DoTheMath 12/4/2012 | 9:34:49 PM
re: A Double Dose of IP QOS A corollary to my argument is that different networks
offering different quality of service levels may indeed
be the economically sensible thing to do - exactly like
McDonalds vs full-service restaurants. Indeed, the continuation
of the voice network separate from the Internet and the
coexistence of land voice vs wireless voice all point to this.

However the cheap and good enough network may get to be so
good that most applications eventually migrate to it, to
take advantage of the cheap pricing, leaving the more expensive
infrastructure stranded. That is beginning to happen in the
wireline to wireless migration, and more speculatively, the
voice to data network migration.

You can see how Southwest is destabilizing the full-service
airlines' business models, to the extent that many of them
may abandon their hub-and-spoke models altogether. Here is a
classic case of "cheap but good enough" class of service crowding
out alternatives. Of course, in the McDonalds case,
the full service restaurants will probably always coexist
with cheap/good enough fast food, but in the airlines case, the customers
are not seeing the extra value.
DoTheMath 12/4/2012 | 9:34:49 PM
re: A Double Dose of IP QOS mrcasual wrote:
Most people find the QoS of their cellphones acceptable
for occaisonal or convenience calls. Most people also
find the QoS of cell phones unacceptable for their
primary business phone to expand on your Bill Gates
analogy.
...
The benefits of QoS can be that a single network
infrastructure (i.e. the Internet) can service many
diverse needs if some form of end to end (and end
to end is a requirement) QoS exists. Without it we
end up with the hodge podge we have today.

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

Agree and disagree. The cell phone vs wireline situation
refers to very different types of networks, and indeed
they are not strictly substitutable (though a lot
of substitution is going on right now). My point is
that within a given network - take the cell phone network -
the operators are not offering services where you pay more
to have your "all important call" go through. By the way
peak hour vs non-peak hour pricing is not the same thing,
because regardless of my ability to pay, I get the
same service as the next guy during each of those
intervals.

Much the same way, McDonalds doesn't offer you a deal to speed up
your service during lunch rush hour, if you pay 10% extra.
My contention is that in most situations, a single service
class (call it "good enough service") is what is economically
feasible. In the McDonalds situation, their goal is to
move people in and out within a set period of time, and during
lunch rush hour they open more counters. Packet networks have
a more than passing resemblance to McDonalds! Any other QoS
solution McDonalds comes up with will a) increase their
cost of doing business b) raise average prices c) piss off
customers and d) reduce their profit or all of the above.

Now, if you want much higher class of service than McDonalds
provides, you have the option of going to a "full service"
restaurant, but then *that* restaurant also offers a single
"good enough" class of service (much higher than McDonalds, but
only one single class of service), uniformly for all patrons.

The point is that in most markets, it is not economically
feasible or sensible to offer differing QoS to different
customers.



lob 12/4/2012 | 9:34:47 PM
re: A Double Dose of IP QOS I completely agree with DoTheMath on the value of QoS to the end-users. (In fact I was a vocal opponent to the whole idea of using switched virtual circuits (ATM and MPLS) for implementing differentiated service for years).

However, I do not think the lack of real interest to QoS translates into the differentiation of networks: delivering packets is just delivering packets - this service does not have the intrinsic differences in the end products (unlike McD's pseudo-food vs real (but more expensive) one). So it is likely to converge on one standard, like the voice networks did.

(BTW, there _is_ a priority voice service for the government use. It has no economical or technoical reason for existance, being simply legistlated).
DoTheMath 12/4/2012 | 9:34:45 PM
re: A Double Dose of IP QOS However, I do not think the lack of real interest to
QoS translates into the differentiation of networks:
delivering packets is just delivering packets - this
service does not have the intrinsic differences in
the end products (unlike McD's pseudo-food vs real
(but more expensive) one). So it is likely to
converge on one standard, like the voice networks
did.
-------------------------------------------------

Agreed. I was just illustrating the point that
if different levels of quality is desired, having separate
networks may in fact be more economical. I am not saying
that is what markets will converge towards. After all,
Southwest gained share in every recession against the
full-service carriers. Initially, the bigger guys though
there is no way business travellers will put up with
Southwest service. A few recessions later, they are not so
sure anymore. Most business travelers nowadays only get
upgraded to business class - they never pay for it - which
is not the model the carriers intended.

I happen to think that the current expensive telecommunications
networks will face the same fate. Even 3G wireless is going
to face the music against cheap and good-enough 802.11. It is
not required that 802.11 networks be "as good" as 3G - if they
are dirt cheap, traffic will move there, knocking out the
economic underpinnings of 3G.
buliwyf 12/4/2012 | 9:34:34 PM
re: A Double Dose of IP QOS Hmm. Agreed with you 100% until the 3G comment.

If you see wireless as simply a substitute for mobile internet (like dial) then yes, 802.11 is 'good enough' (remains to be seen how it scales).
3G (promises) additional services and proprietary content which in turn locks users in to providers and generates more handset sales.

It's a bit like comparing a mud simple internet connection with AOL. Fact is (sadly or not) users like simplicity, universal access and value add content.
Be it 3G or something else, I don't see 802.11 being the dominant mobile protocol.

DoTheMath 12/4/2012 | 9:34:28 PM
re: A Double Dose of IP QOS Hmm. Agreed with you 100% until the 3G comment.

If you see wireless as simply a substitute for mobile internet (like dial) then yes, 802.11 is 'good enough' (remains to be seen how it scales).
3G (promises) additional services and proprietary content which in turn locks users in to providers and generates more handset sales.

It's a bit like comparing a mud simple internet connection with AOL. Fact is (sadly or not) users like simplicity, universal access and value add content.
Be it 3G or something else, I don't see 802.11 being the dominant mobile protocol.

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

Granted. It is pure speculation on my part. But I happen to know
several 802.11 companies, and they are drawing up plans to
take it beyond corporate LANs, and into practically all public places.
Already airports, Starbucks etc have them, and if most malls,
restaurants, etc. have them, you begin to cut into some of the
potential use of 3G. This issue will be decided on how 3G is
priced. If the differential in price between 3G and 802.11 is not
material, yes, 3G does have a shot.

Good point about AOL. Their approach won once they removed
price as the barrier. I remember the stampede that resulted
when they matched the $19.95 pricing then prevalent. Once they got
the user base, they have raised it since, but the user base is slowly
eroding to cheaper services now.
HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE