Learn about Linux
Download Linux
Get Linux helpSpecial Offers
Feature: Open Source
OpenOffice.org seeks P2P technology to help with distribution
By JT Smith
on
June 17, 2002 (8:00:00 AM)
Share
Print
Comments
- By Anne Zieger -
Despite having a healthy list of mirror sites, OpenOffice.org's servers have begun to choke on the demand for binary and source file downloads. After all, the Open Source office suite was downloaded about a million times in May from its official CollabNet servers alone, and no one expects things to slow down much from now on.
Despite having a healthy list of mirror sites, OpenOffice.org's servers have begun to choke on the demand for binary and source file downloads. After all, the Open Source office suite was downloaded about a million times in May from its official CollabNet servers alone, and no one expects things to slow down much from now on.
To make OpenOffice files more accessible -- and give its servers a breather -- community managers are turning to the peer to peer movement, asking members for their best shot at an appropriate P2P file-sharing system.
For the first round, at least, they're handing the project over to another Sun Microsystems-backed community, P2P-oriented Project JXTA. JXTA technology is a set of open protocols allowing connected devices to communicate and collaborate. Members of the JXTA community are working on a variety of consumer and corporate projects, including a handful of content distribution and file-sharing approaches the OpenOffice.org leaders want.
OpenOffice.org has been overwhelmed by the growth in traffic since it released OpenOffice.org 1.0 on May 1, says community manager Louis Suarez-Potts. "When we just had monthly builds it was no big thing, but when we released 1.0 it was a huge deal," he says. "It made it very, very clear that in order for people to get OpenOffice products, we had to have complimentary systems, so it's spread as far as possible."
The server bottleneck should only get worse as the number of files grows -- as users translate the OpenOffice.org suite into their native language, for example -- boosting the demand for downloads, community leaders say.
The idea behind the P2P project is to build on the bandwidth-multiplying benefits P2P file-sharing technology, which should deliver documents more efficiently the more users access a file. Old-style networks clog up when the download frenzy begins, but P2P networks keep adding new nodes, scaling up ad infinitum, proponents say.
What's more, keeping bandwidth requirements to a minimum could help OpenOffice.org participants in foreign countries, who sometimes don't have access to the kind of fat pipes U.S. users typically do.
A well-designed P2P file sharing client isn't a bandwidth hog, notes Sam Hiser, co-lead of the OpenOffice.org marketing project. "You can actually configure your box to limit bandwidth use," he says. "Your participation in a P2P network can be almost invisible. You can deliver relatively few bits compared to what you download."
As it happens, the notion of using P2P to share OpenOffice files isn't new. Frustrated by Web server congestion, OpenOffice.org supporters had already begun to use the P2P file sharing apps like Gnutella to snag copies of the suite.
But OpenOffice.org project managers would like to offer users a more controlled way of getting the latest updates. If a file is distributed across the Gnutella network, there's no way to enforce file integrity or control versioning, notes John Sulski of CollabNet, Project JXTA's community manager.
OpenOffice.org managers are asking the JXTA community to develop a simple GUI-based app allowing P2P discovery, sharing and transfer of requested OpenOffice files in a manner similar to Gnutella or JXTAs' myJXTA peers. Given the bulk of some distributions, they want the system to handle large files easily (up to 100 MB). They're also asking that the app be designed to let users configure the versioning and integrity enforcement on their own, allowing them to download files other than OpenOffice.
Perhaps most importantly, the proposed system will need to check for integrity of the downloads by checking in with known secure peers already storing reference data such as checksums and version information.
Comments
on OpenOffice.org seeks P2P technology to help with distributionNote: Comments are owned by the poster. We are not responsible for their content.
Because the official (master and mirror) sites are (becoming) so slow that using the P2P download option would (hopefully) be faster.
Even with mirroring, servers become the chokepoints for limiting heavy download requirements. A P2P network makes more efficient use of total internet bandwidth by increasing the number of servers. Since most P2P systems implement a "number of hops" limitation, download requests are usually satisfied within a local region of influence and do not bog down the entire internet.
With so many people downloading OpenOffice this is excellent news. I've certainly been promoting it among my friends and contacts.
Since it's open, it will also mean that projects like koffice and gnome office will have a much simpler time creating filters for the OpenOffice native format, making interoperability a dream compared to the Microsoft swamp we are hopefully finally leaving.
Another thought:
I'm not American, but many Americans love linux. Have you considered that Linux will actually hurt you in the end? If countries gain independance from American commercial software, then profits will stay in their country and be kept out of your hands...meaning that your economy won't be so good.
Pro:
* Free software means greater innovation and power in the hands of the people (what really matters)
Con:
* Less jobs for you and your kids, because when there's less money there's less work
Since it's open, it will also mean that projects like koffice and gnome office will have a much simpler time creating filters for the OpenOffice native format, making interoperability a dream compared to the Microsoft swamp we are hopefully finally leaving.
Another thought:
I'm not American, but many Americans love linux. Have you considered that Linux will actually hurt you in the end? If countries gain independance from American commercial software, then profits will stay in their country and be kept out of your hands...meaning that your economy won't be so good.
Pro:
* Free software means greater innovation and power in the hands of the people (what really matters)
Con:
* Less jobs for you and your kids, because when there's less money there's less work
According to basic economics a new business will have to spring up to employ the people that don't have jobs. Many of the people without jobs will develop new things to make and do. They will be able to spend the money they would normally spend on software on more research and development. The beauty of the system is that it can adapt. The technology of the era cannot last forever as has been seen in the industrial revolution now to the computer revolution. We don't know what the next step is but hopefully we'll be able to work with it when it comes.
And people who migrate to US can stay at their homes now, contributing to their growth..like Indians, Chinese, etc :)
As an American I urge you to break the economic and political chains imposed by the American Government and its people.
Band together, opose our intrusions. Big empires like the United States need many lessons in humility. We are an arrogant, proud, greedy, and self centred people.
Open Source will hurt the established industrial complex, but it is for the benefit of humanity and the World as it should be. I will be happy when the United States is reduced to getting by diplomacy what it gets by economic and military might.
Band together, opose our intrusions. Big empires like the United States need many lessons in humility. We are an arrogant, proud, greedy, and self centred people.
Open Source will hurt the established industrial complex, but it is for the benefit of humanity and the World as it should be. I will be happy when the United States is reduced to getting by diplomacy what it gets by economic and military might.
Are you sure you dont work for $Microcrap ??????.
Where the hell do you get all this garbage from...??????????
Where the hell do you get all this garbage from...??????????
Another thought:
I'm not American, but many Americans love linux. Have you considered that Linux will actually hurt you in the end? If countries gain independance from American commercial software, then profits will stay in their country and be kept out of your hands...meaning that your economy won't be so good.
Americans spend more on software, so they will save more. Meantime, commercial interests are outsourceing their programming to India and the like anyhow.
Or perhaps American industries have just as much to gain from the demise of Microsoft as any other in the world?
Cutting software costs means more money for extra employees, better saleries, etc. How much money do you suppose a company such as AT&T could save if they never spent a penny on MS Windows or Office? Probably tens, if not hundreds of millions of dollars, which could be used to employ hundreds or just be used to revitalize departments that need it and improve ultimately improve services.
MS isn't really helping our economy anyway. What little they give back is quickly taken away by strict licensing rules requiring enormous amounts of money to pay for MS software for schools and government offices.
Cutting software costs means more money for extra employees, better saleries, etc. How much money do you suppose a company such as AT&T could save if they never spent a penny on MS Windows or Office? Probably tens, if not hundreds of millions of dollars, which could be used to employ hundreds or just be used to revitalize departments that need it and improve ultimately improve services.
MS isn't really helping our economy anyway. What little they give back is quickly taken away by strict licensing rules requiring enormous amounts of money to pay for MS software for schools and government offices.
Cowabunga!
It's the way of free enterprise - toss the idea out there and let the chips fall where they may. It certainly worked for M$ during the beginning of the desktop computing revolution. This had unfortunate consequences on the rest of us. Now, we do it again - M$ better watch out. Oh and that Gutenburg printing thingy - it'll probably make a few economic changes too.
It's the way of free enterprise - toss the idea out there and let the chips fall where they may. It certainly worked for M$ during the beginning of the desktop computing revolution. This had unfortunate consequences on the rest of us. Now, we do it again - M$ better watch out. Oh and that Gutenburg printing thingy - it'll probably make a few economic changes too.
I am an American and I need to remind you that less than half of us voted for the current executive branch of our government. I hate to bring in politics but just as there are a lot of bad guys in this country, there are many good guys who want to help set the code free and help the people on this planet.
Just a rant, flame away ;-)
Just a rant, flame away ;-)
It's worse then that. Less then half of us who voted, voted for the current executive branch.
Has it ever come to your attention that perhaps all American software companies do not just make Microsoft compatible products.
I used to work for a commercial software company that made & sold a boxed retail product. The main program was a utility that was bolted on to Microsoft's DOS to give DOS some features that nearly every other operating system already had.
Back then, MS was just starting to make an OS extention (Windows 3.x) that required them to delve into the same area. From that point forward, the need for my old company's product shrank to the point that it appears on abandonware sites...mostly as a curiosity.
Am I angry at MS for this? Nope. The old product took advantage of a hole that is assumed filled with any OS you can get today -- free or pay, closed or open.
Keep in mind...
* Most of what is heavily used in the open source world is a commodity or soon will be; operating systems, word processors, databases, ....
* Just because a product is in a commodity category doesn't mean that all nitches of that product are. Some COTS software is worth the fees that they charge -- as consulting, licencing, maintenance, or what have you.
* Most projects are not of the COTS (Commercial Off-The-Shelf aka 'shrinkwrapped') variety. Most work is custom and local.
* COTS expenses are spent on companies where I (and most other US employees) do not work.
* COTS expenses come out of the whole company budget and thus take money from other projects -- including both IT and other software projects.
* Eliminate most of the core COTS software by using open source, and you increase the local money available.
Back then, MS was just starting to make an OS extention (Windows 3.x) that required them to delve into the same area. From that point forward, the need for my old company's product shrank to the point that it appears on abandonware sites...mostly as a curiosity.
Am I angry at MS for this? Nope. The old product took advantage of a hole that is assumed filled with any OS you can get today -- free or pay, closed or open.
Keep in mind...
* Most of what is heavily used in the open source world is a commodity or soon will be; operating systems, word processors, databases, ....
* Just because a product is in a commodity category doesn't mean that all nitches of that product are. Some COTS software is worth the fees that they charge -- as consulting, licencing, maintenance, or what have you.
* Most projects are not of the COTS (Commercial Off-The-Shelf aka 'shrinkwrapped') variety. Most work is custom and local.
* COTS expenses are spent on companies where I (and most other US employees) do not work.
* COTS expenses come out of the whole company budget and thus take money from other projects -- including both IT and other software projects.
* Eliminate most of the core COTS software by using open source, and you increase the local money available.
The software industry is a small portion of the United States economy.
Open source projects like linux can only help the us economy<nobr> <wbr></nobr>... it would be very difficult to download<nobr> <wbr></nobr>;-) and I haven't seen a free open circuit processor or open view monitor or open hard wired dsl or broadband connection<nobr> <wbr></nobr>... free, open software feeds into the need to buy these hardwired realities.
Using p2p as a means of saving bandwidth costs for low budget distribution of a high popularity application is a great idea. But downloading code from unknown sources opens up the possibility of trojans and other nasty surprises. A user can usually trust that a download direct from the vendor or from another known site will be free from tampering. This trust model doesn't work when the download is coming from some unknown, unaccountable peer somewhere on a p2p network. Open source software is particularly susceptible to this - the same access to the source that makes updates and fixes so easy makes inclusion of malicious code easy, too. Inserting trojan code into proprietary binaries requires some work -- usually hacking or outright replacing a library -- so that it still functions anything like the original application. Adding it to open source is far simpler. Sure, the user could crawl the code of everything he downloads, but who wants to do that for apps you just want to use, not study or contribute to? This assumes that the user would even recognize the malicious code if he saw it, which is not guaranteed.
Vendors wanting to distribute via p2p or other uncontrolled mechanism need to create and make available tools to verify the integrity of downloaded software, both source and binaries, against a master list of known good code. With the master list itself being available from known, trusted sources. Using something like MD5, it would be simple for an end-user to quickly check his new download before installing or compiling, and be assured that he wasn't about to hand over the keys to his machine to some bored high school kid in Singapore. And make it simple and child-proof, so that even casual users can take advantage of it. I know that the hardcore command line bigots scorn such ease-of-use niceties, probably wanting everyone to do the checksums in their heads or something similarly silly. Some open source vendors already do this to some degree, making the md5 signatures for ISOs and major releases available for use in some manual checks. But how hard would it be to write some hand-holding gnomified gui something-drake to do the work for the user? Automating tasks is one of the things a computer is for, why not let it do it.
Vendors wanting to distribute via p2p or other uncontrolled mechanism need to create and make available tools to verify the integrity of downloaded software, both source and binaries, against a master list of known good code. With the master list itself being available from known, trusted sources. Using something like MD5, it would be simple for an end-user to quickly check his new download before installing or compiling, and be assured that he wasn't about to hand over the keys to his machine to some bored high school kid in Singapore. And make it simple and child-proof, so that even casual users can take advantage of it. I know that the hardcore command line bigots scorn such ease-of-use niceties, probably wanting everyone to do the checksums in their heads or something similarly silly. Some open source vendors already do this to some degree, making the md5 signatures for ISOs and major releases available for use in some manual checks. But how hard would it be to write some hand-holding gnomified gui something-drake to do the work for the user? Automating tasks is one of the things a computer is for, why not let it do it.
a possible solution to that problem lies in distributed disk drives: have the data broken up in chunks and encrypt these chunks in a way that each host peer does not know what he holds and reassemble the chunks at download.
But the problem is that if I downloaded the source and put in trojan, then put it back to the share folder and let the P2P network share it, even data is encrypted, the Trojan is packed already.
Ximian's Red Carpet does this transparently (unless a signature is wrong or omited in which case you get a message) and is incredibly easy to use.
A similar client for P2P, where you add the keys of 'vendors' you trust would be fantastic. Each vendor could publish signed 'channels' and you could pull signed files from those channels over the P2P network.
That would be, in a word, sweet.
A similar client for P2P, where you add the keys of 'vendors' you trust would be fantastic. Each vendor could publish signed 'channels' and you could pull signed files from those channels over the P2P network.
That would be, in a word, sweet.
Altnet.com is working with the largest P2P network KaZaA. Altnet provides a secure P2P network that puts content owners in control of the online distribution of their content.
This problem demonstrates one of the problems of the free software model. Things cost money. Bandwith, servers, routers, cables, programmers, tech support, etc. You cannot escape this truth. Shifting the load to mirrors may help in the short term, but eventually you have to pay for these things somehow.
Why not take the easy way out and charge $5 a download from your server? I'd gladly pay it to ensure continued access to excellent software.
There is a reason why the free market works, it uses money as the measure of work to take care of problems like these. I'm sure there is a better way to describe this but it all boils down to the old SF adage - TANSTAAFL
There Ain't No Such Thing As A Free Lunch
Why not take the easy way out and charge $5 a download from your server? I'd gladly pay it to ensure continued access to excellent software.
There is a reason why the free market works, it uses money as the measure of work to take care of problems like these. I'm sure there is a better way to describe this but it all boils down to the old SF adage - TANSTAAFL
There Ain't No Such Thing As A Free Lunch
I want to disagree with you. Openoffice.org can use its website for advertising and all. Launch into other services or something. Clearly they have a lot of hits and they could could use advertising revenue to support the site. Besides, it belongs to us therefore we, the users would benefit from it.
The advertisement model just doesn't work. It barely worked during the boom years of the dot-com bubble, and it doesn't even come close now. The money paid by the advertisers has to come from somewhere, and those folks don't have that much money right now to throw around.
Advertising can do much to cover the bandwidth costs for page views for a popular site, but the current payouts don't begin to cover the bandwidth used by downloads of a large application. If the main page is 10k, and the download package is 10MB, then a person getting a copy of the package consumes the bandwidth of 1000 visitors to the page. And the advertiser only pays once, not 1000 times. Do you really believe that an advertiser will pay for each page view one thousand times the cost of serving that page? Not a long term sustainable proposition, not with the current weak advertising prices.
Selling services might help, too, but they would need employees to provide those services, with all of the overhead involved in having workers and invoices and taxes. And what value can the openoffice folks add that someone would want to purchase? The server application authors, such as Apache and Sendmail, can charge for services for installation, integration, customization, and so on. They are dealing with complex software on a relatively small number of machines that most of the users will never see. But an end-user office suite? If it is that tricky to install and use, it won't get chosen in the first place. No service dollars there. Not to say that they can't get any money at all for services---they might do nicely in training and documentation, especially materials designed to ease the transition from Microsoft Office---but the potential market will never be that large, and they would be very lucky to do much more than cover the payroll and other costs of those providing the work.
It's a difficult problem. Open Source may be a mature, technically superior means of software development, but the economics are still in their infancy. Currently it's an unrestrained anarchy (with freeloaders who consume but don't produce) drawing on a barter economy (the open source authors and investors) which is trying to interface with an aggressive capitalism (everyone else, particularly those who actually have money). And it's not working too well. Without a good answer soon, it will be another example for the history books where the technically superior choice lost to the better marketed alternative. Listed right next to Betamax.
Advertising can do much to cover the bandwidth costs for page views for a popular site, but the current payouts don't begin to cover the bandwidth used by downloads of a large application. If the main page is 10k, and the download package is 10MB, then a person getting a copy of the package consumes the bandwidth of 1000 visitors to the page. And the advertiser only pays once, not 1000 times. Do you really believe that an advertiser will pay for each page view one thousand times the cost of serving that page? Not a long term sustainable proposition, not with the current weak advertising prices.
Selling services might help, too, but they would need employees to provide those services, with all of the overhead involved in having workers and invoices and taxes. And what value can the openoffice folks add that someone would want to purchase? The server application authors, such as Apache and Sendmail, can charge for services for installation, integration, customization, and so on. They are dealing with complex software on a relatively small number of machines that most of the users will never see. But an end-user office suite? If it is that tricky to install and use, it won't get chosen in the first place. No service dollars there. Not to say that they can't get any money at all for services---they might do nicely in training and documentation, especially materials designed to ease the transition from Microsoft Office---but the potential market will never be that large, and they would be very lucky to do much more than cover the payroll and other costs of those providing the work.
It's a difficult problem. Open Source may be a mature, technically superior means of software development, but the economics are still in their infancy. Currently it's an unrestrained anarchy (with freeloaders who consume but don't produce) drawing on a barter economy (the open source authors and investors) which is trying to interface with an aggressive capitalism (everyone else, particularly those who actually have money). And it's not working too well. Without a good answer soon, it will be another example for the history books where the technically superior choice lost to the better marketed alternative. Listed right next to Betamax.
If you're so keen to pay for it why don't you donate? It's basicly the same thing except it means that there is no pay system to complicate things.
Wouldn't work. Someone would download it, and put up a mirror to share it for free because they would consider that to be "noble".
And then not only would you have mirrors anyway, but you would have mirrors that aren't trustworthy.
And then not only would you have mirrors anyway, but you would have mirrors that aren't trustworthy.
But, we already know download charges or more properly charges for download time doesn't work. That was the financial model of the old online services--CompuServe, Delphi, AOL, GEnie, etc. Once the Web became popular, no one was willing to pay for downloads, or the time to make downloads. Only AOL is still around in a significant manner and they do it as a flat rate ISP spending marketing dollars like a dry drunk in a bar. A few sites still try something like it, eMusic charges $9.95 to download from their small, but legal, MP3 collection. But, generally speaking, no one makes money charging for downloads. Of course, no one tends to make money from the Web anyway, but that's another matter! ;-)
Steven
Steven
Why charge for downloads? A P2P network works perfect. The load is removed from the servers and a signiture verifies the packages. Chunks of the packages can come from 20 or 100 or 1000 different sources, get peiced together and verified. It is a wonderful parallel to the open source development model!
By uploading once a day to Usenet, every news server in the world becomes a mirror. User connections to their ISPs news server is typically very fast (at or need line speed).
And the side benefit is all the extra unused bandwith on the Internet gets absorbed. 100% utilization is a good thing, right?
Seriously why send it to timbuktu if no one there is ever going to download it? And then repeat every day.... I suppose it would help reduce the porn in newsgroups because the length of time everything is spooled would have to be decreased....
Seriously why send it to timbuktu if no one there is ever going to download it? And then repeat every day.... I suppose it would help reduce the porn in newsgroups because the length of time everything is spooled would have to be decreased....
Very nice system..MIT License, works with IE and Mozilla, fast etc
http://www.bitconjurer.org/BitTorrent/
http://www.bitconjurer.org/BitTorrent/
Why P2P
Posted by: Anonymous Coward on June 17, 2002 08:46 PM#