Re: ISPs whinging about BBC's iPlayer



"Cork Soaker" <ISawYourMotherLast@xxxxxxxxxxxxx> wrote in message
news:ftqn99$dse$1@xxxxxxxxxxxxxxxxxxxxxxxxxxxx
: A cache only does any good for the 2nd access or for multiple
simultaneous
: accesses where you can tie them together and do something about it.
:
: the killer for any caching scheme is bandwidth reduction vs cost of
building
: and operating the cache.
:
: So given 5000 broadband exchanges in the UK, how many copies would you
"use"
: of the same item at a specific exchange before you purge it?
:
: Even the biggest online TV style events only attract sub 1 million
viewers,
: which means there you have efficiencies of 100 to one on a good day, and
on
: demand content must be less - just because there is so much more choice.
:
: I suspect the cache "efficiency" is only 2 to 10 times for typical
content.

You are right, but a reduction is a reduction.

yes - but not worth doing unless the up side is a gain compared to the
complications.

:
: Then - can you share the cache between the different ISPs? And LLU
operators
: have separate plumbing - so that fragments the user set even more.

Of course you can, this is how P2P works. It's all about spreading the
data
around. It's not *the* solution, it's just something else that might make
a
difference. Content providers have reported a significant reduction in
bandwidth when switching to this kind of technology, including the BBC.
Do
you want to bet that the ISPs that these providers use aren't moaning
about
the change in bandwidth? A REAL test is needed....
Comcast in the states have teamed up with BitTorrent using DNA technology.
If it didn't work, they wouldn't bother.
Comcast provide Internet services to end-users and content providers, so
if
they're saying they are seeing an improvement, you better believe them!

:
: then you have to buy the servers, run them, maintain them when they go
: wrong, pay BT for rack space and power and so on.

The obvious problem here is the monopoly.
A simple system can be a cache, especially when it comes to P2P. You
don't
need complicated systems, and it's only there to help reduce bandwidth.
The
software is free and the ISPs swap out older equipment all the time - they
may as well use it!

the s/w might be free - but the dominant cost for evan an existing server is
running / maintaining / power & other services and eventually replacing it
when it dies or you cannot get spares anymore.

Private-owned networks like cable companies would see the obvious
immediate
benefit.
BT themselves provide Internet access, so they should see it too.

:
: no - caching isnt going to work without a fundamental change to the ISP
/
: exchange plumbing.
:
: > IPv6 allows for this kind of transition too, where IPv4 would have
failed
: > (or struggled).
:
: AFAICT this is irrelevent - there is nothing at the network level that
IPv6
: can do that you cannot get out of IPv4.

Not really, the IPv4 address space is limited. There's no way the world
could handle extra devices at the exchange running IPv4, and NAT is not an
option here. IPv6 is designed with ridiculous numbers in-mind. It's only
a
simple issue, but it's been resolved so it may as well be used.

yes - limited. but still the raw address space has around 2000 million
useful addresses.

although it is limited, there is still fair bit around - and most ISPs will
have spare space in their blocks.

After all - 5000 cache servers isnt much compared to the monthly growth of a
broadband ISP and they find at least 1 address per consumer for that....
:
: > The data travelling back down to the ISP backbone doesn't cost them a
: penny
: > though - they don't have to pay for their own traffic! So they won't
see
: > any reduction in cost by doing this - so it's of no interest to them.
:
: the assumption here is that an ISP own their own backhaul.
:
: If they use BT IPstream then they dont, and there are usage charges - so
not
: in that case.
:
: Even if they are LLU based, many backhaul links seem to use rented
fibres
/
: lambdas or Ethernet pipes - so more capacity is fine as long as the pipe
: isnt full, but at some point there will be a step change in cost.
:
: Spread over lots of links to lots of exchanges that will follow the same
: cost curve as incremental cost for the extra bandwidth - even on their
own
: fibre the equipment needed to drive a link at 10G is more expensive than
1G
: and so on.
:
: Even if raw bandwidth costs were zero, it takes a lot of IP and
transmission
: hardware to distribute IP plumbing across the UK - significant capacity
: upgrades will make Cisco / Juniper / Alcatel et al very happy.......

True. But it's not a recurring cost. The extra purchase exists to reduce
bandwidth, so should last.... (famous last words).


old rule of thumb was whatever price you get the hardware for, maint came in
at around 10% / year.

then uplift the hardware to allow for install. Then pay for the Netlocate @
BT exchanges to put them in (or if it is BT, justify not being able to sell
that space to someone).

:
: cable - no idea, but VM keep getting slammed for having bottlenecks on
their
: network.

Yes, well, they are still offering staggering speeds compared to DSL, but
that's a technology thing...
They would benefit the most from redesigning their network, less "core"
more
"regional".

: >
: > But, surely, THE ISPS KNOW ALL OF THIS. They're meant to be experts
in
: the
: > field! They just want more money. If they're not happy with the ISP
: > business, they can pull out! The Internet was NEVER meant to be
privately
: > owned in the first place, it will evolve whether they want it to or
not.
: > They know this too.
: > Grab the Pound notes while they still can.
: >
: > Oh yeah, and the BBC is NOT the first do this by a long shot. They're
one
: > of the last to jump on the bandwagon.
:
: Agreed. And people like INUK have a business model based on IP TV as
their
: main product....
: http://www.inuknetworks.com/

It's sad, but a lot of people think that the BBC are the first to do this,
because it was in the news! These are the same people that believe
everything they see on TV of course, and watch Big Brother ;-)

:
: > The ISPs are screaming at the BBC because they have a better chance of
: > squeezing money out of them, because it's public money and nobody
really
: > looks after that!
:
: now that bit i believe....

It's fairly clear though isn't it?

:
: >
: > The commerical channels would just tell them to bugger off, and they
know
: > this also.
: --
: Regards
:
: stephen_hope@xxxxxxxxxxxx - replace xyz with ntl
:


A response with some intelligence! Thanks Stephen :-)
--
Regards

stephen_hope@xxxxxxxxxxxx - replace xyz with ntl


.



Relevant Pages

  • Re: ISPs whinging about BBCs iPlayer
    ... : and operating the cache. ... Then - can you share the cache between the different ISPs? ... bandwidth when switching to this kind of technology, ... :> The data travelling back down to the ISP backbone doesn't cost them a ...
    (uk.telecom.broadband)
  • Re: Some Kudos for Dell
    ... These come at a cost that most providers would rather avoid. ... > to provide remote support we had to get SBC to allow us to use Bridge ... expects his ISP to support whatever hardware he has, ... >> well as the network servers and the network it's self. ...
    (alt.sys.pc-clone.dell)
  • Re: Some Kudos for Dell
    ... >>> to provide remote support we had to get SBC to allow us to use Bridge ... >>> mode in order to get a public IP so that we could manage the NAT on ... > no cost to the user or the ISP. ... >> question came up about providing at cost Antivirus software. ...
    (alt.sys.pc-clone.dell)
  • Re: The newbies dilemma!
    ... monthly fee to an ISP for Broadband connection to the Internet. ... wouldn't be able to justify the cost of keeping them if I did. ... PS Lots of advertising for your programme in this thread, eh, Dustin? ...
    (microsoft.public.security.virus)
  • Re: DIY space transport
    ... >>>Isp is over rated. ... It's cost per pound that matters, ... >>>payload per pound of vehicle. ... >>>engine was really quite depressing as far as Isp... ...
    (sci.space.policy)