Peters: The Desktop or the Browser: Is the Netbook Escalating the Battle?

Posted Aug 10, 2009 21:34 UTC (Mon) by wmf (guest, #33791)
In reply to: Peters: The Desktop or the Browser: Is the Netbook Escalating the Battle? by drag
Parent article: Peters: The Desktop or the Browser: Is the Netbook Escalating the Battle?

I think IPv6 is useful and P2P is really cool, but they have nothing to do with Web 2.0 replacing desktop apps. When someone switches from Thunderbird to Gmail or iPhoto to Flickr, that's all client-server. IPv6 would make no difference. And the hottest thing in VoIP is Skype - a P2P system.

Peters: The Desktop or the Browser: Is the Netbook Escalating the Battle?

Posted Aug 11, 2009 2:58 UTC (Tue) by dannyobrien (subscriber, #25583) [Link] (6 responses)

Yes, but the reason most Web 2.0 services are client-server is because people want their content either available on any computer they use without configuration hassles, or shareable with their friends. Both of those are issues with reachability -- you could do both by having your desktop or a home server do it, but it's currently more trouble than it's worth to do that. Easier to have Google cope with the problem.

The question is: is that a feature of a centralised, third-party controlled service, or something that autonomous, free software-running computers can compete with?

It's a bit like the "can Linux compete on the desktop" writ large. Can the Linux ecosystem compete with features Google, which presents an easy-to-use, always-available, set of features with just a dash of lost privacy, inflexibility, lack of source code access, and network lock-in?

Peters: The Desktop or the Browser: Is the Netbook Escalating the Battle?

Posted Aug 11, 2009 3:29 UTC (Tue) by dlang (guest, #313) [Link] (5 responses)

you seem to be forgetting that most of the web 2.0 services are running on linux, so there's no question that linux can compete there.

there are a lot of reasons besides NAT/IPv6 that people do not run such systems out of their houses. maintaining an always-available system is non-trivial, both in knowledge, maintinance effort, and money. it's hard to do at home even for those of us who do this sort of thing routinely at work.

you aren't really wanting to know if linux can compete, you are wanting to know if opensource applications with an AGPL-type license (that forces the code to be available to anyone who uses the website) can compete.

in general I think the answer is probably not.

everything is getting turned into a commodity from the low-end up. first you had the hardware (the PC), now linux is doing the same thing to the kernel and OS level, but there is always something custom somewhere in the stack. once in a while it's special hardware with a fairly generic user laer on top, but usually it's the specific software that you are wanting to use that is the tailored part of things.

specific applications are going to be commoditized, I expect that (if ODF ever takes hold) the word processor will quickly fall into this category (like the text editor it grew up from did), and I would also expect the spreadsheet category to go that way but I expect that there will continue to be other new 'must have' packages that show up.

Peters: The Desktop or the Browser: Is the Netbook Escalating the Battle?

Posted Aug 11, 2009 4:26 UTC (Tue) by dannyobrien (subscriber, #25583) [Link] (4 responses)

I agree with you that competing with managed services "in the cloud" is hard, because one of their advantages is that they *are* managed by professionals. But there are also advantages to having one's own control over software, and that control has traditionally been one of the reason why users and administrators have chosen open source offerings over proprietary offerings.

I think that there's more to the Linux ecosystem than just what's running underneath Google, and more options for it to survive than simply attempting to oblige Google and other centralized companies to use AGPL licensing (perhaps I erred in emphasising the Linux connection, but this is LWN after all. Let's call it the FLOSS ecosystem). And commoditization can work both ways. Many of the advantages that Google originally had in the marketplace and which they leveraged to create their new products have now been commoditized: Hadoop, distributed file systems, innovation in the Javascript space. Google's core competancy, search, still seems (to me at least) to be something that a centralised company can excel at. But I don't see why a Flickr or GMail service has to, per se, depend on a single company, a single server farm, or me handing over all my data to a third party.

Peters: The Desktop or the Browser: Is the Netbook Escalating the Battle?

Posted Aug 11, 2009 8:04 UTC (Tue) by mjthayer (guest, #39183) [Link]

E-Mail is always going to be a special case, but for other applications, I think that a Linux desktop combined with storage in the cloud is excellently well placed to compete. The weak side of web applications is the difficulty of migrating to a different service. However a typical Linux desktop contains the tools for dealing with most interesting types of data out of the box, so if it can painlessly connect to the network storage, the user is just as free to switch to a different physical system and work as they are used too. Of course, if the system is controlled by someone else, this presents a greater privacy risk than "pure" cloud services :)

It's matter of scale...

Posted Aug 11, 2009 10:40 UTC (Tue) by khim (subscriber, #9252) [Link] (2 responses)

But there are also advantages to having one's own control over software, and that control has traditionally been one of the reason why users and administrators have chosen open source offerings over proprietary offerings.

Yup. And this one percent of user will continue to use free software and AGPL applications. But we are talking about remaining 99% here: will they continue to use Windows and spend billions for upgrades, AV software and support, or will they use something else?

Most users don't care (never cared and will never care in the future) about control (that's why Kindle is so hot). They don't want control - they want something usable and the less hassle the better!

But I don't see why a Flickr or GMail service has to, per se, depend on a single company, a single server farm, or me handing over all my data to a third party.

Because selling point of Flickr or GMail are not the stellar software behind the offering, but actual service. You know: servers, administrators, that kind of things. And these benefit from "economy of scale" enormously. Software is of secondary concern here.

It's matter of scale...

Posted Aug 11, 2009 12:46 UTC (Tue) by NAR (subscriber, #1313) [Link] (1 responses)

Most users don't care (never cared and will never care in the future) about control

Exactly. And don't forget that in order to control, one needs specific training, skill, probably even some talent. Not everyone has that and most people are not willing to invest in that. Driving a car is so much simpler that administering a home server, yet look around in the streets that how many people are driving a car who really shouldn't, because they lack the necessary skills.

It's matter of scale...

Posted Aug 11, 2009 13:58 UTC (Tue) by drag (guest, #31333) [Link]

Driving a car is a lot harder then it seems. :) Most people take about 4-5 years before they are technically proficient at driving.

Anyways....

-------------------------------------

Maintaining a server at home is difficult because Linux servers are targeting professionals. This does not mean that setting up a simple P2P server has to be difficult, though.

Am I going to think that 'apt-get install apache2' and then editing text conifguration files can be made easy for the 'average person'... NO! Of course not. But that does not mean that a 'desktop internet share' service should be difficult or require lots of software.

All you need is a small python-based http server and some other protocols to setup and your done. It's not going to be much different from the stuff it's going to take to make Gnome and KDE be 'online' and 'connected' with online Web 2.0 stuff. Write it in high-level code and design it for simplicity and the security problems should be rather small.

The only 'server-client' relationship that is fundamentally needed is one used to find other people on the internet. So you need to have a server that a person can connect to in order to find other people. DNS is good for servers, but getting a DNS name is fairly expensive and since most people have a combination of a NAT firewall + dynamic IP addresses it makes it fairly worthless nowadays to find individuals. So you have to find something else...

And that "something else", in my eyes, is XMPP.

It doesn't take much. There are free servers all over the place.

Using XMPP you get online notifications, friends listing, and you can locate people on the internet, blah blah blah. Pretty much everything you'll need. You don't need to have people typing in IP addresses or whatnot. And if the person's desktop is offline then you are not going to waste your time trying to access them or look for them.

If a person is walking around with their laptop and whatnot and connecting to wifi hotspots or traveling then you'll be able to find them easily, too. Once they connect then their desktop should send connect to their XMPP server and all their online friends will know about it. If they don't want to be bothered then a single button on a Applet would be all they need to go into 'shadow mode'

Because you different notifications right? Offline vs away and all that stuff. Offline means that you can't access the content. Away just means that the person is away from the desktop, but you should be able to access the content served by the desktop if you have authorization.

So I don't think it's impractical to use XMPP as a location and notification service and then use that to locate people and initiate file sharing, give authorization for viewing personal data, twitter-style mass-messaging, chatrooms, photo albums, mass file sharing via bittorrent, and whatnot.

and it's all using Kerberos and whatnot so you never have people sending passwords and whatnot over the internet as a way to identify themselves.

Besides IM and file sharing then other stuff like VoIP and music streaming should just fall into place naturally.

If it's kept as simple as possible and the installation is easy and requires no configuration beyond getting a online identity, then it should actually likely be easier then going to all these different online places and setting up accounts.

And if people want to tie all of this into online services then that should be possible, too.

Then the difficult thing after that is going to be how to transverse firewalls, which most NAT firewalls are actually surprisingly insecure and easy to transverse. You can use techniques like uPNP to request port forwarding on most routers sold in the past few years, for example. You just have to look at the P2P folks and look at the techniques they use to get data to each other. (Transmission Gnome bittorrent client does a good job, it seems)

----------------------

I had a experience about how difficult it is to setup a 'server' on the internet a few years ago.

I was visiting my folks and my brothers were there. So I decided to setup a little 'LAN Gaming' action with Nexuiz.

It was fun. We were playing and then all of a sudden I thought my brothers changed their username. I made fun of him and somehow that person did not appreciate my humor all of a sudden.

Turns out that Nexuiz has uPNP support so that when you setup a gaming server, even if it's behind your firewall, it opens a port and sends out notifications to the internet 'Master' servers.

I had, completely by accident, setup a internet gaming server and people were joining and playing against me and my brothers without me realizing it right away!

---------------------

I know that right now my Gnome Desktop has some stuff for sharing files and whatnot. But it's way too heavy and requires setting up Apache and whatnot for Webdav.

That's the wrong approach.

For example for LAN sharing a simple http/dav python server on a high numbered port can send out Avahi or uPNP notifications that other desktops can pick up on is all you need. No need to setup some heavy multithreaded server for that sort of thing.

It does not even really need to be running the entire time. Having a desktop equivalent to a Inetd super-server is all you need.