Rodi (P2P)
How to...
Message board

SourceForge.net Logo
Get Firefox

Valid HTML 4.0! Valid CSS!

Posts http://larytet.sourceforge.net/posts/posts.r

This forum is closed for posting. Please, use http://sourceforge.net/projects/larytet/.
Also you can ask questions on board.planetpeer.de P2PForums.com IntegrityP2P.com
Thank you.

Try to refresh the page to post

Name: larytet E-mail:
Thu, 2 Mar 2006 20:26:40 -0700

Generate interrupt when FIFO moves from empty to not-empty

I am not sure if this idea is patented. I post it here to make a case in the future on prior art.

Imagine a device which adds events (for example, data packets) to a FIFO in shared memory. To improve overall performanec of the system such device can generate an interrupt when FIFO state moves from empty to not empty. The following algorithm is suggested

- if valid bit at tail is set drop the event
- add event to the tail and set valid bit in the entry
- if valid bit in the entry (tail-1) is 0 (cleared) generate interrupt (level interrupt)
- increment tail

- ack the interrupt
- loop
- if valid bit at head is set handle the event
- get out of the loop
- clear valid bit at head
- increment head
- back to loop
- return from interrupt

Name: larytet E-mail:
Thu, 24 Feb 2006 20:26:40 -0700

Peer2Mail, Gmail Drive, G2G Share

This looks as a neat idea to make use of large e-mail accounts for data distribution.

Publisher chops file by multiple chunks, uploads the data to multiple e-mail accounts and inform the subscribers that the data is available. Subscriber wishing to get a copy notifies the publisher. Subscriber sends his/her e-mail to the publisher. Publisher can require from the suscribers full access to the e-mail account - publisher will use the account to offload his own e-mail accounts. Publisher forwards the data to the e-mail account of the subscriber. Subscriber pulls the data from the e-mail account.

What is interesting here is that the publisher does not allow access to the 'original' e-mail account. The file can not be removed easily. It is not easy to trace the file back to the publisher. Internet e-mail provider can not distinguish the e-mail accounts of the publisher by 'heavy' usage pattern. From publisher and subscriber point of view delivery takes zero time - in the best case this is time required for copying the data from one account to another on the same server.

Publisher can sign for POP3 access (usually costs money) and this way to make sure that the delivery does not fail because of changes in the HTML of the Internet e-mail engine.

The easiest way to solve problem of subscription is to use some anonymous e-mail account, like one at MyTrashMail.com, or IRC channel. Publisher can check anoymous e-mail account for e-mails of new subscribers or run bot to collect IRC messages.

In similar way can be solved problem of 'index server'. IRC channel or multiple anonymous e-mails can serve as a list of releases.

Publisher can employ different tactics for hiding the identity (IP address). For example, publisher can use multiple e-mail accounts and chain them. Publisher sends the data to e-mail account B from account A. After that publisher access account B using regular HTTP proxy to serve requests from the subscribers - forward the e-mail to the subscribers after removing header with IP address.

The only real problem of the system is the ever changing look-and-feel of the Internet based e-mails. In some cases Internet e-mail provider can ask heavy users to 'enter the string you see on the picture'. The only way to fight this is using multiple e-mail accounts and POP3 access.

Another interesting option is hiding the data in the image files. Services for keeping photos are gaining popularity. It is not going to be too hard to generate PNG file containing 128K chunk of data. Typically such services limit the total size of the files stored and in some cases the total traffic. Multiple accounts will solve the problem.

It's also a good idea to add script engine for writing plugins and ability to exchange plugins. Adding a new WEB site or modifying of the exsisting plugin should be easy procedure. Probably a good reason to learn Phyton ?

BTW if the application described above gets really popular ads based services will start to feel some pain.

Name: E-mail:
Wed, 20 Apr 2005 17:12:17 -0700

Self destructing hard disk

Device prodution cost is expected to be under $100 and retail price is in the area of $200. spec is in work.

Some highlights
- device includes motion sensor with sensitivity controlled by SW
- device works with and without external power supply

Disk destruction (combustion) can be initiated by
- motion sensor inside of the box
- external sensor
- power down and/or power up event
- remotely using daemon (agent) running on the machine

Two modes of work are supported
- Enable/disable from keyboard or remotely
- One shoot only
- Scheduled explosion

Mode 'one shoot only' is when user can only enable the device. In this mode the device once installed and configured can not be disarmed.

User will attach the device to the hard disk and connect serial or parallel connectors to configure the device.

Combustion will destroy magnetic surface of the hard disk and sensitive elements on the mother board, like SRAM and NVRAM chips.

send email to larytet@yahoo.com if you want to be enlisted as a future customer

Name: larytet E-mail:
Thu, 14 Apr 2005 15:51:44 -0700
from Official eMule-Board
Jul 27 2003, 04:17 AM

The EMULE developers are proud that they have JUST NOW produced a release that PROVES your identitiy to all other clients by SECURE and IN-CIRCUMVENTABLE cryptographic algorithms.

Name: larytet E-mail:
Thu, 17 Mar 2005 10:01:28 -0800
originally posted on Yahoo Bittorrent message board

i wonder again and again why one would use 4Kb blocks. i chose 4M (mega) blocks - full bit-map of blocks for 8GB file can be sent in a single Ethernet packet.

let\'s forget for a second about overhead BT creates. The faster peers learn who has what the better overall performance is going to be.

Loosely related issue. Let me doubt the reasons behind choosing of bencoding format, instead of compressed XML or JASON if are looking for something less \"bloated\". I think it was a simple choise, because it was ready to use in Phyton. or i am wrong ? Let\'s save one day of coding and think about the cost later. The cost though is not marginal, far from it. Overhead of the protocol is a linear function of a number of sent packets which, because of small block size, is a linear fucntion of the file size. i think that in reality the function is stronger than linear, because in the situation where we have a large file with only one seed and large swarm most of the traffic is generated BT clients agressively looking for the blocks.

Let\'s be honest. BT is created for 0.5-1G files and for the network of multiple hosting servers (seeds) interconnected with fast links. 10G+ files is not exactly the BT \"market\".

And what about lossless video codecs ? 24 frames/s * 1024 * 1280 * 60 * 60 gives us 100G/hour. with 1:4 compression 25GB/hour. Two hour video clip is 50G. are we going to use 4KB blocks for this kind of files too. and calculate hash tree using mainframe ?

Name: larytet E-mail:
Thu, 24 Feb 2005 09:55:45 -0800
...In an effort to keep pace with the evolving Internet needs of our Rogers Yahoo! Hi-Speed Internet customers and to continue providing you with a fast and efficient service, we have implemented a combined upload and download bandwidth limit of 61,440 Megabytes (60 gigabytes) per month. This is a very generous limit which is approximately equal to the consumption of a typical customer in an entire year ...

Buy unlimited business connection (fiber or a couple of E1s) and use WiFi to deliver service to 5-10 families to divide costs Essentially i suggest to run ISP. it is goiing to be after all more expensive than regular cable, but you will get what you want 5MB symmetric 24/7 connection assuming that your subscribers belong to 98% consuming 60G/year I don't see this hurting them in the near future. They will at most lose the 2% of people that go over this because of file sharing but they will free up a much higher percentage of their bandwidth for their regular users.

It can hurt, because without filesharing and video on demand nobody needs banhwidth above 512K and that inlcudes 4 additonal phone lines. me thinks that ISPs are preparing to serve their own VoD from their own video servers. this of course is going to be free of bandwidth caps. Comcast today makes money on 'adult' content. nobody makes real money on the data providing. video services is real gold mine. the only problem guys have that the content is too expensive. they can't simply send you movie without paying dues to the you know who.

what can also be done is filesharing behind the same hub or installing your own LANs/WiFi hot spots. with advanced antenas (like a can of cola) wifi can deliver data up to 5km range. this is illegal in most countries, but it is not easy thing to catch you. your backbones are Wifi streams (cola cans), your close range connections are regular LAN - cable or 4 wires.

Once somebody downloaded popular video from that moment local hub is used.

... and let me remind you about Rodi project - dedicated for the highspeed networks with low packet loss rates (LAN). This is also first file sharing application which in the functional requirements contain IP ranges and not IP addresses. means that you can prioritize what IP addresses you download/upload from first.

Name: larytet E-mail:
Thu, 24 Feb 2005 09:48:32 -0800
I am looking for a program that is private but also includes multisource downloading. I tried reverse connect but my friend coulnd't index his families somethign was wrong. My upload speed is 40 k/Bs and thats not fast enough. Is there a WORKING program that works semi like bit torrent, wheres its more of a community *PRIVATE* Would a torrent tracker work? Is that really private? I just want to be with a small group of friends.

Rodi provides you with ability to download from mulitple sources and does not require any installation. tell your friends what is your IP range and port number and run the application or applet directly from the website. that's it. Last version supports chat see http://larytet.sourceforge.net/btRat.shtml

...and no problems with dynamic IP - you tell your friends the range of IP addresses your ISP uses for your hub.

the application is developed with performance in mind. in the regular test i run three Rodi clients on the same PC with total traffic 300Kbytes/s and still get aroud 0% CPU usage. size of the binary is 250Kbytes. yes, you read it right - quarter of megabyte. and most important - this is open source. if you want you can download the code, make required changes, compile (batch files compiling the code on Windows are available).

Name: larytet E-mail:
Sat, 19 Feb 2005 11:19:21 -0800
From email: ...Further, most of them (individuals sued) may also have hard drives filled with other evidence of infringement (you'd be amazed what ... forensics guys can dig up)

for Windows read, edit if required and run batch file bigRedButton.bat

This file will remove sensitive materials and fill the disk with some pattern. It takes time and can even crash your disk, but the disk is going to be erased in the process better than after format command.

The rabbit-hole went straight on like a tunnel for some way, and then dipped suddenly down, so suddenly ...

Name: larytet E-mail:
Mon, 31 Jan 2005 13:38:07 -0800
List of Statuses looks wrong for me.
Not really. Application will not always connect callback, but can poll the status too. at least it was the idea. i need latest state of the session and different End of Session statuses is a way to forward error code. alternatively i need additional method to get session error code.

You dont provide neither DT statistics structure nor methods to extract it
i am not sure what kind of statistics data layer can provide and in what form. DataTransfer is plugin and eveloped by somebody else and can use different approach to the statistics. i am tryig to keep requirements as basic as possible

In all place where you mention Packet Processor call it Application PP
This is PacketProcessor for the application and this interface is application side requirements. your plugin can have any PacketProcessor it needs but this is in separate code and probably not even in Java

Constants for UPLOAD and DOWNLOAD are both 0s - I dont think you meant that.
this is a bug obviously

The name naive is useless - better stop and go or somethink like this.
i would prefer to keep stop and go for some other plugin

If I understand your intentions properly, uploader doesnt need to reduce rate - its rate is dictated by downloader
No, not at all. DataTransfer plugin should make an effort to max up available physical link(s) over multiple connections. Optionally DataTransfer plugin can support field maxBandwidth in the child of the class DataTransferSessionParameters. I think there is a misunderstanding what plugin is in my view. Plugin is a library which is loosely integrated with the main application.

Name: princejer E-mail:
Mon, 31 Jan 2005 10:40:27 -0800
Below please find my notes to Data Transfer protocol.

  1. List of Statuses looks wrong for me. Basically, you mix up status and event. As soon as DT layer has been stopped its status must be IDLE or something like this. The reason of stop better be event, reported to application or saved in history.
  2. You don\'t provide neither DT statistics structure nor methods to extract it.
  3. In all place where you mention Packet Processor call it \"Application PP\". It\'s not beyond the possiblity that DT will use its own instance of PP class with its own set of IEs. I even think this will be the only reasonable design for any more or less complex DT protocol (like UDT).
  4. Constants for UPLOAD and DOWNLOAD are both 0s - I don\'t think you meant that.
  5. There can be many other parameters for session in future, but I deem a failure timeout is a mandatory one even for a naive protocol.
  6. The name \"naive\" is useless - better \"stop and go\" or somethink like this.
  7. If I understand your intentions properly, uploader doesn\'t need to reduce rate - its rate is dictated by downloader.

Name: princejer E-mail:
Thu, 20 Jan 2005 11:39:02 -0800
Looks like you need much more buffers for sync requests than for async ones.
The reason is that sync buffer will be kept by the request manager until an answer will arrive while an async buffer may be freed right away after sending.
So, roundtrip delay and downstream capacity will influence sync pool size as well.

Name: princejer E-mail:
Wed, 19 Jan 2005 18:45:15 -0800
File version should be introduced as one of search fields along with its name, description, etc... in view to support Version Manger.

Name: princejer E-mail:
Wed, 19 Jan 2005 18:42:19 -0800
Request Manager\'s UML class diagram is wrong - sockets (both RX and TX) are not its chldren. I suppose, they might be in association with the Manager.

Name: princejer E-mail:
Wed, 19 Jan 2005 18:38:05 -0800
The little endian example in design has annoying typos - better fix them do not confuse readers.

Name: princejer E-mail:
Wed, 19 Jan 2005 18:33:32 -0800
IE size field must be two bytes because 254 bytes IE data may not be enough for XML and Management IE elements.

Name: E-mail:
Sat, 8 Jan 2005 19:11:41 -0800
from Content and Control: Assessing the Impact of Policy Choices on Potential Online Business Models in the Music and Film Industries
...it (Internet) may enable content creators to act altogether independently of the traditional intermediaries
(this PDF is must read in full btw)
Open Source Movies ? and actually why not. reasonble equipment cost under 10K and there are plenty of people dreaming about acting. Link from the same article Hacking the Broadcast Flag with the Digital Television Liberation Front
also from MPAA, FCC, more from EFF, pcHDTV

Open source "Build HDTV recorder by yourself" or "HDTV recorder in 3 easy steps" would screw the whole idea. Some guys at FCC did not think enough. There is nothing in the receiver which can not be reproduced from widely available parts. I can imagine electronic company specializing on these parts and making good money.

For example, if the industries rely on high sales during the limited initial period of impeded piracy, this approach might intensify the industries' current tendency to invest in a few immensely profitable hits while marginalizing the vast majority of content creators. Generating high initial sales will require large upfront costs for marketing and the technological impediments to piracy, and, in this way, this approach may make risk-spreading intermediaries like the record labels and film studios more valuable. In this way, the benefits of the dropping cost structure and greater independence of content creators might diminish.
In other words strengthening of monopoly of RIAA is not good for artists or for vast majority of artists.

Name: larytet E-mail:
Wed, 5 Jan 2005 00:57:27 -0800
is it able for a 3rd party to produce a list of all members participating in p2p sharing
The short answer is yes. If adversary runs server (adversary is a publisher or content provider) it can collect IP addresses of all peers (downloaders). Rodi does not provide 100% anonymity. There is a long answer though which is yes and no. Adversary can not reliably prove that logged IP adresses indeed belong to the senders, because downloader hiding behind bouncer can argue that the IP address was spoofed/faked. There is another problem for the adversary - it\'s own IP address. Typically downloader will look for the content on the crawlers and crawlers will accept links to the content only from reliable/friendly source. Bottom line is that Rodi first of all protects the publishers and only then downloaders. Publisher needs bouncers because of possible DDoS attacks. Partial anonymity is a second issue and natural outcome of using connectionless protocol.

Name: Hypnotoad E-mail: Hypnotoad86@gmail.com
Tue, 4 Jan 2005 21:43:53 -0800
Aussie is just someone from an irc where i was talking about this. If you can delete those posts that is much appreciated. You\'ll have to excuse my security incompetence but if the website running the Rodi java client is encrypted than is it able for a 3rd party to produce a list of all members participating in p2p sharing. And since security is such a major issue towards all p2p products produced how can one defend themselves from the likes of netstat -n?

Name: larytet E-mail:
Thu, 30 Dec 2004 10:03:42 -0800
To Shaw Fiberlink users: I assume that the Ellacoya promise that the shaper is statefull is true. In simple terms it means that they create an entry for every established TCP connection and keep the record in the database thorugh the whole lifecycle of the connection (i have my doubts about how reliable such thing can be if they do not terminate the connection continue

Name: larytet E-mail: larytet@yahoo.com
Wed, 29 Dec 2004 18:02:09 -0800
From now every post comes with link
Printble format http://larytet.sourceforge.net/posts/posts.html

Name: larytet E-mail:
Wed, 29 Dec 2004 12:44:58 -0800
To Lucas
Could somebody explain in layman\\\'s terms how packet shaping works?
see http://www.p-cube.com/products/serviceEngine.shtml
in simple terms i can put a box between your modem and IP router - so called middle man. IP router is the box with default gateway IP address. For example,

Windows IP Configuration
Ethernet adapter Wireless Network Connection 2:
        Media State . . . . . . . . . . . : Media disconnected
Ethernet adapter Local Area Connection:
        Connection-specific DNS Suffix  . : XXXXX
        IP Address. . . . . . . . . . . . :
        Subnet Mask . . . . . . . . . . . :
        Default Gateway . . . . . . . . . :
in this case default gateway is a box with IP address and my PC IP address is this middle man box is transparent for you, but it can do lot of things. For example, it can follow TCP SYN request (connection request you send to any remote peer to establish TCP session). The box can record all tcp connections you do and trace the packets going through the connection in both directions. The box can recognize BitTorrent/eMule message headers, because they are well known and simply drop the packets to limit byte rate/packet rate for this specific TCP connection. Regular HTTP/FTP connection will work perfectly, because the box knows to recognize HTTP packets.
From p-Cube website
  • Layer 7-3 stateful wire-speed packet inspection and classification
  • Robust support for over 600 protocol/applications including:
  • General: HTTP, HTTPS, FTP, TELNET, NNTP, SMTP, POP3, IMAP, WAP, & others
  • P2P: FastTrack-KazaA, Gnutella, WinMX, Winny, Hotline, eDonkey, DirectConnect, Piolet, and others.
  • Streaming & Multimedia: RTSP, SIP, HTTP-STREAMING, RTP/RTCP, etc etc.
  • Programmable system core for flexible reporting & bandwidth control
  • Transparent network and BSS/OSS integration into existing networks
  • Subscriber awareness for relating traffic and usage to specific customers
  • Traffic redirection for HTTP and RTSP applications
ISP buying this equipment only because of P2P traffic shaping are wasting their money (and my money too). There is nothing more easy than to fake regulat HTTP packet or SIP message.
Is there any way to get around traffic shaping at this particular time?
Use regular FTP servers. Use http://www.yousendit.com/ and publish on some message board Use P2P networks supporting encryption, like Mute, Ants, Tor and many others. Use new P2P networks which is not recognized by the middle man. Use VPN between you and the peer - requires VPN client installed on both PCs. Proxy servers should solve the problem too. Ask Azureus/eMule team to make changes in the TCP payload (message header) to make it similar to HTTP GET request. Pay attention that this change is not going to be backward compatible. It can be done in the way that new clients will work with old clients. Let\'s say that client N is a new cient and client O is an old client. O trys to connect to N. N recognize that the recieved packet is not HTTP. N knows from that that the peer (o) runs old software and will send old style packets. etc etc. i think you can make conditional donation to eMule/BT/other. talk to them. bring this question to their forum on SF.net. For most of the traffic shapers workaround is going to be easy. In case of toughest traffic shapers weak encryption will help. for example, sender can XOR the payload with URL string, torrent file contents or file hash. This way the key is known to all peers and unique for every file. no middle man can collect all keys and decrypt the payload.
Call your ISP and require from them to stop traffic shaping. Move to ISP which does not limit the bandwidth. Send link to this page to your ISP and tell them that they are wasting your and their time and money.
Install FTP server (there are plenty of free/shareware/open source) on your desktop. Publish your IP address or use NoIp service. This way you can share everything you want on your disk. Your ISP probably does not shape FTP traffic. Istall HTTP server it works exactly the same. Use Coral network http://www.scs.cs.nyu.edu/coral/overview/ to improve performance.

Name: Lucas E-mail:
Wed, 29 Dec 2004 10:33:37 -0800
Is there any way to get around traffic shaping at this particular time? Could somebody explain in layman\'s terms how packet shaping works? I don\'t understand how the switches pick out the P2P traffic without limiting the regular internet traffic.

Name: larytet E-mail:
Fri, 24 Dec 2004 22:30:03 -0800
unrelated udea. imagine regular torrent with swarm 10,000+ peers. If average number of connections is 50 the probability for a IP address to be logged by adversary is 0.5% or 1:200. The easiest path to anonymity in existing BT networks is scale. Or one can say that BT is not secure BECAUSE it is NOT popular enough. One can also argue that BT is not centralized enough. Instead of only one torrent file with latest TV show there are 3 or 4 running simultaneously in different formats, codecs, on different trackers. If BT network breaks 50 mil users my wild guess that most popular items will be relatively secure. BT tracker can divide the swarm into groups with 50-100 participants. When a new peer arrives tracker can join the peer to already existing group or create new group and move some seeds/leachers from the existing groups. Chance that adversary will connect/see IP addresses belong to more than one group are

where n is a number of IP different addresses belong to adversary and G is a number of groups The chance for any peer to be logged is the same. Tracker can make intelligent guess based on the subnet that multiple IP addresses belong actually to the same peer and move addresses to the same group while removing not related IP addresses. Interesting side effect of this would be that geographically close peers (sitting behind same DHCP server) would find each other in the same group. All this is relevant only for large swarms on the order of number of available ports/connections on the regular desktop. We are talking about 10-50K participants. RSS feeds containing links to torrent files and automatic download tools based on these feeds can create such swarm.

Name: larytet E-mail:
Tue, 21 Dec 2004 23:09:58 -0800
From now on posts are in reversed order - newest posts come first.

Name: larytet E-mail:
Tue, 21 Dec 2004 21:45:53 -0800
to DeadMan: there is notion of crawlers. Downloader can search crawlers instead of scaning the whole network. WEB sites are not out of the game either - they simply optional. Publisher can list IP address(s) of it\'s own server or IPs of friendly bouncers. Crawler can have regular WEB interface too and require from publishers to go through registration, etc. Both public and private crawlers can coexist running on the same server.
The idea with downloaders driven rate is interesting but it was done before and found unreliable in case of decentralized networks. The main problem where and who is going to keep this score (see http://www.gnu.org/software/gnunet/ for more info)
Size of single post is limited by 500 chars or smething like this. This is not my server and I do not have much time to moderate the posts and the server space is limited and ... etc. etc. there is also 3 sec delay from post to post. It gives me some time to clean up any mess created by PHP before SourceForge kicks me out.
You do not have to specify your nick name and/or email. I personally promise that there is absolutely no any log of IPs in the PHP code supporting this message board. Still SF.net can collect IPs accessing this or other pages, but to my best knowledge there is no such log.
Did you read the main page of the project or just comments on other websites ?

Name: DeadMan E-mail: no@noway.com
Tue, 21 Dec 2004 21:03:45 -0800
...got cut off. He downloads the file or previews the file and discovers the file is not what it proclaims to be. He can now add a rating and/or comment to the file saying the file is fake, low quality or whatever. The more people who rate it the redder a rating icon gets until it gets blown off the network entirely. This feature would have to be secure enough not to be used by adversaries as a means to expire legitimate files of course.

Name: DeadMan E-mail: no@noway.com
Tue, 21 Dec 2004 21:01:52 -0800
I\'ve been reading about this project and have an idea to add. The idea for the protocol sounds good. However a system like bittorrent relied on torrent links being added to websites as verified links and was moderated. In an open P2P system there is always the potential for fake files to be served regardless of the security of the protocol itself. What is needed is a way to supply verified trustworthy links to files which can be moderated. One way of doing this while steering clear of a centralizing index and also avoiding having someone moderate it is to put the onus of moderation onto the users. This can be done via a rating scheme. A user searches for a file. The file shows up as only being shared by a few people. He joins the download/upload queue (BTW it would be nice to be able to preview certain content that can be previewed such as video and music). He downloads the file or previews

Name: larytet E-mail: larytet@yahoo.com
Mon, 20 Dec 2004 14:17:41 -0800
Rodi will NOT give 100% anonymity, though publishers will find it relatively easy to use spoofed IP addresses and friendly bouncers. Indeed the filexibility is first priority goal of the project. I am trying to define protocol as open as possible. The whole idea of information elements is there to make the protocol easily extendable. i\'ll do my best. actually i am behind my mental \"scheduler\" by about a month. hopefully before end of the month (Dec, 2004) you will have an ability to search files on the remote peer (which is usefull by itself) and in the end of Jan 2005 you will get working download - only simplest, without swarming, traffic shaping and control packets bouncing. For all this time the application is going to remain command line only. it will take half an year to reach the point where performance is comparable with existing BT clients.

Name: Stealth E-mail: stealthgod2002@yahoo.com
Fri, 17 Dec 2004 18:21:06 -0800
I think that BitTorrent is almost at the end of its lifespan. It has become too high-profile and now ISP\'s and organizations like the RIAA and MPAA are looking to bring and end to it. In my case, my ISP is using packet shaping to drop the majority of the upload packets, which in turn slows download speeds. Of all the BT alternatives I\'ve seen, I think this is the most promising and could be a lasting solution. I\'ve read over your design and I really like how you\'re keeping things very open and flexible (as BT is) so that other developers will be able to build on your foundations. How long do you think it will take until we start seeing Rodi in action? Keep up the good work!

Name: arkady E-mail: larytet@yahoo.com
Sun, 5 Dec 2004 09:24:24 -0800
hello william
you are the first person who used my "mesage board". thank you. apparently you found a bug there, because i did not receive email notificatiion from SF.net and the script was supposed to send email.
are you a native English speaker ? my English grammar is really bad. i could send you HMTL files to correct the most obvious mistakes
regardig traffic shaping. i think that eventually retail will pay per byte like businesses pay. it means that basically the bill will look like electricity bill - monthly flat fee + fee depending on amount of consumed power. let's say $20/month infrastructure (includes first 1Gb u/d) + $0.01/Mbyte above 1Gb.
if you need low latency, low delay, low jitter services for VoIP, video, etc. you will buy such links separately.
existing traffic analyzers can be fought rather easy if, for example, BitTorrent starts to use weak encryption. Let's say that part of the torrent file is a 128 bytes key which used to XOR with all sent and received packets. In this situation traffic analyzer can only use traffic statistics to find "P2P" peers and traffic analyzer will limit the whole traffic not only P2P so it will be obvious and clearly seen by ISP client. Try to post mesage on Azureus message board. they can make the change rather quick just to check that it helps.
For Rodi this is only part of the idea. The major idea is using UDP (connectionless) protocol. It will provide better line utilization, faster download times for fat links (like 10M+) and some degree of anonymity for the publishers, because publisher can always argue that UDP packet in the log was faked. The only way to prove that publisher sent any specific packet is to log the whole traffic from the publisher, which is not always easy (ISP can be in different country) and in many countries requires permission of judge.
Another strong part of Rodi is content search. think about distributed Google. no single point of failure. any desktop is a server storing searchable information. multiple crawlers facilitate the search process. this is going to be huge (if happens).

Name: William E-mail: william@bcwebnet.com
Sun, 5 Dec 2004 01:24:27 -0800
I haven\'t completely read through you\'re website but it appears Rodi may solve \"traffic shaping\" issues. Specifically with d/l & u/l of BitTorrents for me. Unfortunately I\'m no porgrammer, nor a qualified enough PM or Tech Writer but if you need a tester, someone to bounce ideas off or someone to do research let me know. I\'m online 9am to 9pm - most days of the week.

Name: arkady E-mail: larytet@yahoo.com
Sun, 28 Nov 2004 22:14:07 -0800
my first experiment with PHP after 2 days of reading of manual at PHP.net
good manual btw, thanks
...and yes, you can leave message here

End of posts

Release 0.4 (source) GPL

Message board