Thursday, April 24. 2008The Reality Of P2P Constraints By Australian ISPsJohn Linton An Exetel employee sent me this link yesterday: http://torrentfreak.com//images/vuze-pl ... esults.pdf as it amused him to see what seems to be the 'outing' of several 'high profile' Australian ISP executives as having lied to the industry and their customers for some considerable time about whether or not they treat P2P traffic differently to other traffic on their parts of the network. If you've been involved with, or an observer of, Exetel for any length of time you may recall the torrent of abuse and derision heaped upon our small company when we publicly announced that the rapid, and increasing growth in P2P traffic had to be addressed by us as it was 'swamping' our available bandwidth at peak times - mainly at night on week days between 8 pm and 1 am. This was mainly on third party forums and predominantly 'authored' by the psychologically challenged fanatical supporters of other ISPs. However much of the 'forensic' attack was written, usually anonymously, by ISP employees and in several cases by very senior ISP management. The reasons for Exetel using technology to identify P2P traffic and restrict the bandwidth made available to those protocols were simple, blindingly obvious and absolutely commercially necessary at the time we introduced them. They were, in fact, so blindingly obvious and commercially necessary that I made the comment at the time that EVERY ISP in Australia would now do the same as Exetel planned if, in fact, they had not already done it. ....and, of course other ISPs had already done so and over the succeeding 12 months EVERY ISP in Australia that had not done so followed Exetel's 'lead' and bought the same equipment that Exetel 'pioneered' in Australia or equipment from other similar companies. ALL other ISPs continued to deny that they were, or ever would, do what Exetel had publicly stated it would do and made condescending comments about "little' ISPs that 'didn't have the money to provision a network properly" - and much worse than that thereby encouraging a more widespread encouragement of written abuse in the various on line and even print media. It never concerned me as it did no damage to Exetel's short or medium term business and I took great pleasure in the public squirming and verbal 'back pedaling' that the various other ISPs who had so vehemently denied using P2P controls when they were exposed as pointless liars by events and the evidence of their own user's. An extraction of Australian ISPs from the article referenced above gives the following: ISP: Median Reset Rate AAPT: 20.23% Westnet: 13.86% Adam Internet: 13.33% Internode: 12.24% Exetel: 11.45% Telstra Bigpond: 11.31% IInet: 10.21% Netspace: 10.07% TPG: 9.98% If this information is to be believed, and I'm not sure that it is correct (I'm not technical enough to know one way or the other but I actually don't believe that Exetel uses the sort of processes described in the article) then the ISP who made the most 'attacks' on Exetel about P2P controls and who persistently went to great pains to publicly state that "we do not constrain P2P traffic in any way on our network" INTERNODE appears, from this article's analysis, to apply greater constraints than Exetel does. I have no idea whether that's true or not but what I do know for a fact is (based on seeing the equipment in their racks at Data Centre's Exetel employees visited) is that several the ISPs who were so emphatically denying that they were constraining P2P traffic had installed either the same manufacturer's equipment as Exetel had done or equipment form a similar manufacturer. Does it really matter? Not remotely from my point of view...... .....though you do have to wonder why so many, apparently senior managers within Australian ISPs, seem to believe they can and should lie so obviously about something that will, sooner or later, demonstrate them to be untrustworthy public spokesmen for their companies. Ironically, as no-one who so publicly criticised Exetel bothered to understand from the announcement on P2P controls made by Exetel (and I have to confess that I did try and write that announcement in a way that the information may have been missed by the more strident idiots who, for reasons known only to themselves, seem to only exist to make derogatory remarks about our small company) P2P controls were simply a first, and transient, phase of dealing with the effects of P2P on network management and provisioning. Since that time Exetel has put in place the second phase (P2P caching) and will complete the third phase by the end of this year. In the mean time the amount of 'constraint' on P2P traffic continues to decrease month by month, at least as implemented by Exetel, and in any event the actual implementations over the time we've had them in place have made only minimal differences in download speeds to the vast majority of P2P users of the Exetel services. There will continue to be challenges in adapting networks to handle multi-thread processes which are just one more challenge to be handled by any organisation that provides network based services. Trackbacks
Trackback specific URI for this entry
No Trackbacks
Comments
Display comments as
(Linear | Threaded)
>actual implementations over the time we've had them in place have made only minimal differences in download speeds to the vast majority of P2P users of the Exetel services.<
my meager use/experiences with p2p was so vastly improved by these implementations, i just thought i'd thank you guys again. so.. thanks! Comment (1)
No ISP can afford not to reign in P2P usage. The only ISP that may not use any constraints, network management or provisioning of this sort of traffic would be those that offer an extremely low download limit in both peak and off peak times. Their users probably wouldn't use P2P for obvious reasons.
But I doubt that sort of ISP would be very popular. I use a bit of P2P/BitTorrent. But certainly don't abuse the technology, or the generous downloads that Exetel offer. Comment (1)
By giving credence to this report, I think Exetel are showing their lack of technical prowess. TCP RSTs are part of TCP, and they occur for reasons other than the use of magical p2p shaping boxes.
If an instance of BitTorrent connects to a host no longer participating in a torrent, and there is no firewall in place, a TCP RST will be sent back to the client. This happens a LOT as hosts join and leave a torrent. Using TCP RSTs as a metric to determine if a given ISP is shaping p2p is utterly absurd. This report has zero credibility, as it is obviously created by people who have no idea about TCP. By continuing to hype this, Exetel have lost credibility. Regards, Dan (Network person not affiliated with any of the named providers in any way) Comments (2)
I have no technical expertise - perhaps you noted that comment about my level of knowledge in the post you are referring to.
There is no 'hype' - just a comment that irrespective of what is said, and as pointed out by another comment prior to yours - any ISP that doesn't control P2P usage over its network is commercialy insane. Exetel has enough technical expertise to have built a communications business providing services to more than 70,000 customers from zero to whatever it is today. I note the owner of the IP address you posted from and the 'tenor' of previous remarks from that IP range. Why bother to read my comments? Comments (3)
Regardless of your own technical expertise, you see it fit to assert that engineers from competing ISPs are liars - based purely on the flawed data in this report.
I know for a fact, that at least some of the named ISPs who are alleged to employ traffic control over P2P simply do not. Regarding your comment about my IP address, I think you are incorrect, as I have never posted to this blog before, and have probably only visited this site once or twice - at most. FWIW, I'm not attacking Exetel's business model. In fact, i do admire Exetel's business model - you guys appear to be quite successful. This post was linked from another forum, of which I am not a member, and I found it curious that Exetel were in a position to comment on other providers networks. Regards, Dan Comments (2)
"Regardless of your own technical expertise, you see it fit to assert that engineers from competing ISPs are liars - based purely on the flawed data in this report."
You really must learn to read more carefully if you are going to bother to read at all. 1) I pointed out that report ascribed processes to Exetel that Exetel don't use. 2) I commented that I had no basis for believing the report actually showed anything of real interest. 3) I said my knowledge that some ISP managements had lied was based ON DIRECT OBSERVATION OF P2P CONTROL BOXES being instaled in their racks. 4) I then said IT DIDN'T MATTER IN ANY WAY AT ALL because P2P had to be controlled by EVERY network owner and NO network owner would incur the enormous costs of NOT doing it. I have no idea why you bother to read my comments but if you do and then chooose to comment in future please read the whole of the comment. Comments (3)
The tool used to record the RSTs was somewhat "flawed" in that it counted all RSTs regardless of where they came from, not just RSTs related to P2P traffic.
Comment (1)
None of which had anything to do with what I posted - exept tangentially.
I had already said it was reporting on processes not used by Exetel and any correlation it drew was meaningless. The point was that many of the Australian ISPs who claimed not to use P2P controls have already been demonstrated to have lied about their claims. The whole point was that P2P controls, used by every network management, are necessary and it's pointless to attempt to retend they aren't. Exetel, a tiny ISP, has already moved beyond that initial process - I'm sure many others will too. I note that you post from a similar IP range as "tom". Comments (3)
Limiting and caching P2P I agree is coimpletely necessary. I just hope Australian ISP's don't follow Rogers in Canada and start limiting Skype and VoIP traffic. Lets hope non-P2P "Net Neutrality" remains both in Australia and worldwide.
Yes I know Skype uses P2P but it is not the type you are referring to. Comment (1)
|
Calendar
QuicksearchArchivesCategoriesBlog AdministrationExternal PHP Application |