P2P Caching

Old, inactive threads
Locked
cryptwizard
Posts: 105
Joined: Tue Sep 28, 2004 8:16 am

Post by cryptwizard » Sun Dec 02, 2007 6:48 am

Definitely has an impact when I download from public trackers, it maxes my line speed. Doesn't seem that good for private trackers though...

kingy
Posts: 35
Joined: Sun Aug 08, 2004 7:56 pm
Location: Byron Bay - North NSW

Post by kingy » Sun Dec 02, 2007 1:00 pm

CoreyPlover wrote: The "$10 more" rider to your existing plan dealt with a second option that was (and probably still is) under consideration, and that was $10 for 50Gb per month of cached traffic in addition to your existing peak + off-peak allowance.
Thanks CoreyPlover,
I wonder what Exetel would do after the 50gb Cached Content Allowance was Reached?
Charge for Excess Cached Content? e.g. $0.20 per GB?

Cheers

Munka
Posts: 289
Joined: Sat Oct 22, 2005 8:22 pm
Location: Rural NSW

Post by Munka » Sun Dec 02, 2007 1:23 pm

Well my bumping around on public trackers has done me no good at all, after trying several torrents I have not hit the cache.
I guess those old episodes of the Beverly hillbillies just isn't everyones cup of tea :D

On Hel$tra ADSL1 out in the sticks NSW 220.233.xxx.xxx, if anything it may be a bit slower for me, at least.
Munka

Howard
Posts: 41
Joined: Sat Jan 07, 2006 10:07 am

Post by Howard » Sun Dec 02, 2007 5:58 pm

I got full speed as well. ADSL2 powertel sync at 4308kbps, download at 430KB/s

Dave_10492
Posts: 14
Joined: Tue Jul 17, 2007 7:27 pm
Location: Sydney

Post by Dave_10492 » Mon Dec 03, 2007 9:28 am

Hi Guys,
I'm in NSW on the 220.233.xxx IP range, and noticed a massive increase in p2p speeds in the last few days. Pulling through 130KB/s traffic on my 1.5mbit connection most of the time :D.

I've got a couple of questions, both in regard to the IP's .etc of the caching system.

The first is based off a post I read on Whirlpool. I'll post it here in it's entirety:
"It's early days yet, but I'm wondering if heavy P2P users might yet find a sting in the tail of the Ultraband 2000 caching devices - as in seeding becoming difficult or perhaps even impossible (in some circumstances).

When a (Exetel) seeder begins to upload a file "the layer 7 redirector transparently redirects P2P traffic to the Ultraband cache" (ref: the PeerApp website description).

Once the cache acquires the file, any "outside Exetel" leecher requesting the file is redirected to the U2000. At the same time, the U2000 will make a connection to the "inside Exetel" seeder in order to "maintain transparency" ie. to fool the BT client into thinking there's a "normal" peer connected.

However the external leecher (and any others following later) then downloads the file entirely from the U2000, with no data supplied from the Exetel seeder.

When the Exetel seeders BT client next updates the tracker, the BT client will report no seeding has taken place, so it seems to me that ratio will suffer.

Perhaps there is an offset, the tracker might not be aware of some downloads. If this is so, it remains to be seen whether the two effects would cancel each other out.

I've been unable to find a detailed operational description of the U2000 anywhere on the internet. However it sure looks like the U2000 is capable of mimicking any computer anywhere and generating counterfeit packets supposedly from that computer.

The packet counterfeiting done by the U2000 seems to employ similar technology to the Sandvine devices that stop Comcast seeders from seeding to peers outside the Comcast network.

If the U2000 is capable of emulating Sandvine functions, it would present a very attractive option for Exetel management to do the same as Comcast. Yes, it's a big "if" ... Comcast have much deeper pockets than Exetel so perhaps the U2000 offering is much less versatile than the Sandvine device.

Just my 2 cents." - WP user Harry Kiri

What do you guys think. Once the cache gets a hold of a file, does this mean that no exetel users will be able to seed it anymore?


2nd Question, which is related to the first, and may well answer it. When I'm downloading from the cache, does all traffic appear from one IP (that of the cache), or does it appear from multiple IP's (where the traffic originated from I assume)? As I said, I'm getting 130KB/s combined, and even 140KB/s+ one one torrent when it's the only one running. The thing is, this traffic is appearing from multiple IP's. One this morning was getting 50KB/s from one IP, and 30KB/s from 3 others. Is this the effects of the cache, or just a massive coincidence?

NetworkAdmin
Posts: 559
Joined: Tue Jan 06, 2004 1:19 am
Contact:

Post by NetworkAdmin » Mon Dec 03, 2007 9:38 am

Dave_10492 wrote:What do you guys think. Once the cache gets a hold of a file, does this mean that no exetel users will be able to seed it anymore?
I haven't noticed any problem, but there is one sure way to find out - seed a file and see.
2nd Question, which is related to the first, and may well answer it. When I'm downloading from the cache, does all traffic appear from one IP (that of the cache), or does it appear from multiple IP's (where the traffic originated from I assume)? As I said, I'm getting 130KB/s combined, and even 140KB/s+ one one torrent when it's the only one running. The thing is, this traffic is appearing from multiple IP's. One this morning was getting 50KB/s from one IP, and 30KB/s from 3 others. Is this the effects of the cache, or just a massive coincidence?
1. What were the IP's?

2. Does it matter?

CoreyPlover
Volunteer Site Admin
Posts: 5922
Joined: Sat Nov 04, 2006 2:24 pm
Location: Melbourne, VIC

Post by CoreyPlover » Mon Dec 03, 2007 9:53 am

Dave_10492 wrote:What do you guys think. Once the cache gets a hold of a file, does this mean that no exetel users will be able to seed it anymore?
Harry Kiri's argument does seem correct. I have noticed that my share ratio plummets because of the high speed of direct cache hits. The speed and download stats do indeed look like they are attributed to a random user. This user is also not necessarily an Exetel user because the cache intercepts all requests for packets, even those to international seeds.

But it is still possibly to seed torrents because non-Exetel users that participate in the swarm will still need to connect to a non-cached source. So the only reduction in seeding comes from between 2 Exetel users which is quite negligible.

abdul
Posts: 110
Joined: Tue May 17, 2005 8:57 am
Location: SYDNEY
Contact:

Post by abdul » Mon Dec 03, 2007 10:02 am

Hi All,

We've cached just over 1.4Tbytes of data so far. Our tests have indicated a significant increase in P2P speed mainly.

Below is a screen capture indicating the traffic (66.82Mbps) from the cache server. The way it's calculated is (Cache-out - Cache-in) to get the traffic value of 66.82Mbps.



Image

Dave_10492
Posts: 14
Joined: Tue Jul 17, 2007 7:27 pm
Location: Sydney

Post by Dave_10492 » Mon Dec 03, 2007 10:09 am

NetworkAdmin wrote:
I haven't noticed any problem, but there is one sure way to find out - seed a file and see.
I'll give it a shot :).
1. What were the IP's?
Here's a few off a popular torrent:
85.72.118.109 - 65KB/s
67.35.113.117 - 39KB/s
128.39.42.230 - 29.5KB/s

It's funny how they're all at 99.9% completion.

Here's a screenshot: http://tinyurl.com/3am747
2. Does it matter?
I guess it doesn't, I'm just interested to know how the cache works, that's all.

NetworkAdmin
Posts: 559
Joined: Tue Jan 06, 2004 1:19 am
Contact:

Post by NetworkAdmin » Mon Dec 03, 2007 10:12 am

Dave_10492 wrote:
1. What were the IP's?
Here's a few off a popular torrent:
85.72.118.109 - 65KB/s
67.35.113.117 - 39KB/s
128.39.42.230 - 29.5KB/s

Here's a screenshot: http://tinyurl.com/3am747
They are external, so 'real' sources of the file you are getting.

Dave_10492
Posts: 14
Joined: Tue Jul 17, 2007 7:27 pm
Location: Sydney

Post by Dave_10492 » Mon Dec 03, 2007 10:15 am

NetworkAdmin wrote:
Dave_10492 wrote:
They are external, so 'real' sources of the file you are getting.
Ah, ok. What else have you guys done to the p2p network then? If I'm not benefiting from the effects of the cache, something else must have changed. I have never got above 50KB/s on a torrent from a public tracker before.

tocpcs
Posts: 523
Joined: Sun Aug 26, 2007 10:01 am
Location: Online

Post by tocpcs » Mon Dec 03, 2007 10:17 am

Dave_10492 wrote:
NetworkAdmin wrote:
Dave_10492 wrote: Ah, ok. What else have you guys done to the p2p network then? If I'm not benefiting from the effects of the cache, something else must have changed. I have never got above 50KB/s on a torrent from a public tracker before.
1. The reduction in menace level P2P traffic has freed room on the links for your activity to go through?

2. The IPs are still being reported as where it was received from, despite coming from the cache?

Dave_10492
Posts: 14
Joined: Tue Jul 17, 2007 7:27 pm
Location: Sydney

Post by Dave_10492 » Mon Dec 03, 2007 10:19 am

tocpcs wrote:
1. The reduction in menace level P2P traffic has freed room on the links for your activity to go through?

2. The IPs are still being reported as where it was received from, despite coming from the cache?
Two good points there, any input on these NetworkAdmin?

NetworkAdmin
Posts: 559
Joined: Tue Jan 06, 2004 1:19 am
Contact:

Post by NetworkAdmin » Mon Dec 03, 2007 10:22 am

tocpcs wrote:
Dave_10492 wrote:
NetworkAdmin wrote: 1. The reduction in menace level P2P traffic has freed room on the links for your activity to go through?

2. The IPs are still being reported as where it was received from, despite coming from the cache?
1. No, the traffic comes from the cache, but there must be a real source.

2. Yes.

tocpcs
Posts: 523
Joined: Sun Aug 26, 2007 10:01 am
Location: Online

Post by tocpcs » Mon Dec 03, 2007 11:03 am

NetworkAdmin wrote:
tocpcs wrote:
Dave_10492 wrote: 1. No, the traffic comes from the cache, but there must be a real source.

2. Yes.
You disagree with a reduction in P2P that is cached is allowing other activities such as uncached P2P to also speed up due to removing congestion from the link that the netenforcer would otherwise shape?

Locked