News for hackers. Stuff that matters.
Re:Decentralized Twitter? You just don't get it. by mbell on Saturday August 18, 2012 @08:43PM (#4402815)
I get the impression the author doesn't have a clue what 'realtime' means. I appreciate he is trying to say 'within seconds' but that is not realtime, true realtime is a completely different game.
Re:Decentralized Twitter? You just don't get it. by protomyth on Saturday August 18, 2012 @09:04PM (#4402862)
Yep, twitter is not the classical realtime taught in school. He probably would have done better by referring to it as "everyone in the same living room commenting on the TV". Its timeliness is directly linked to how many people are at / watching the same event. The tweets about a NASCAR race or the plane coming down in the Hudson really would be a pain / confusing if received hours later. For a lot of stuff, this doesn't matter so much (like e-mail).

"Pause and Consider" is pretty much a lost concept in the online news world. Look at the Apple screw fiasco for a point when the repeaters didn't. Maybe having a distributed network of sorta timely servers wouldn't be such a bad thing. I seem to remember debates happening fairly rapidly on USENET.

Re:Decentralized Twitter? You just don't get it. by mbell on Saturday August 18, 2012 @09:26PM (#4402911)
> Maybe having a distributed network of sorta timely servers wouldn't be such a bad thing. I seem to remember debates happening fairly rapidly on USENET.

I believe this is what twitter does. At least that is the impression I have as a user. Having 'watched' a number of live events on twitter I tend to get a floods of 10-30 tweets at a time all in the same language. I presume this means there is a delay related to geographic servers packing up multiple tweet for replication.

Re:Decentralized Twitter? You just don't get it. by thinkingisfun on Saturday August 18, 2012 @09:23PM (#4402907)
There ought to be a way to bring IRC into all this..
Re:Decentralized Twitter? You just don't get it. by ChiperSoft on Monday August 20, 2012 @01:23AM (#4406392)
Exactly, IRC networks have been realtime AND decentralized for decades, they even solved the redundant data problem.
Re:Decentralized Twitter? You just don't get it. by volaski on Saturday August 18, 2012 @10:45PM (#4403052)
Why do you get that impression? I do have a clear understanding of realtime in the sense you are implying. That kind of "realtime" system is actually closer to the point I'm trying to make. Generally centralized architecture is better for building robust and reliable systems than decentralized.

In fact it would be great if you could share some counter examples.

Re:Decentralized Twitter? You just don't get it. by mbell on Saturday August 18, 2012 @11:12PM (#4403094)
'Realtime' may have many definitions but those I'm familiar with relate to near instantaneous feedback based control systems. More generally it applies to systems with mandatory and very limited time constraints. A common high level software example would be algorithmic stock trading. A more historically proper 'realtime' system is code/hardware that put MSL rover on the ground on Mars. These are not things which twitter does. Their UI may give the impression of 'realtime', their systems however are not, nor do they even pretend to be realtime systems.

In short, you've taken a term and misappropriated it based on marketing hype. That is a reasonable thing to do when pitching to a bunch of consumers/marketing folks, it doesn't usually set all that well with engineers.

Re:Decentralized Twitter? You just don't get it. by volaski on Saturday August 18, 2012 @11:25PM (#4403111)
Like you said, "realtime" has many definitions. And I disagree it's a misappropriated term based on marketing hype. Twitter IS realtime, just in a different context. If you stick to your strict definition, it would be incorrect to even call things like XMPP realtime.
Re:Decentralized Twitter? You just don't get it. by mbell on Saturday August 18, 2012 @11:42PM (#4403141)
I have no idea how your bringing XMPP into this but just as Node, AMQP, ActiveMQ, ZeroMQ, Jetty, or <insert messaging or async tool here> they have have no singular impact on a system being realtime or not, they are just tools. If you want to consider a system which involves these tools and has a realtime constraint e.g. X arrives before Y time or we're seriously screwed, then sure, I'll happily concede and call the system, not the tool, realtime.
Re:Decentralized Twitter? You just don't get it. by smlacy on Saturday August 18, 2012 @11:40PM (#4403136)
Clearly, you've never worked in distributed systems, and your impression of Twitter (and other large "centralized" by your definition systems) are actually huge, loosely coupled decentralized systems.

The real difference is whether than decentralized, loosely coupled system is run by and owned by a single corporation, or not.

It's the design of the system itself that determines latency, not the centralized vs. decentralized argument.

You're speaking as if twitter.com is a single giant machine on a single IP address. In fact, it's thousands of loosely coupled machines and services all working in tandem across nearly the entire globe. Sounds like a typical decentralized system to me.

Re:Decentralized Twitter? You just don't get it. by volaski on Sunday August 19, 2012 @02:55AM (#4403429)
I have a feeling that everyone in this particular branch of the thread seems to criticize me based on their disagreement with the definition of terms I used.

I understand it can be confusing because I used these terms interchangeably as well as in more loose context, and I apologize for being incoherent. I tried to elaborate my logic throughout the rest of this thread here and there, so if you have time, please read through them.

Trust me. I do know a thing or two about the technologies I mentioned in the post.