Introduction to Hookbox

[Michael Carter]

Greetings,

As some of you may know, I've been working on an alternate comet/websocket/web-push server called Hookbox (http://hookbox.org) for the past year. The goal of the project is to expose much higher-level apis than Orbited, and provide tight integration with existing web technologies, like PHP, Django, Rails, Java Servlets, etc.

Please take a look at an introductory article/tutorial that I just published on cometdaily: http://cometdaily.com/2010/07/26/a-fast-introduction-to-hookbox/ . I hope that you'll find the technology to be *much* more approachable than anything else out there. Please let me know if you have any feedback.

Thanks,

-Michael Carter

[leon]

Michael,

I took a look at the documentation and would like to ask whether you
plan on handling the message broker within Hookbox itself. I am
currently using Orbited and on the production server I went with
ActiveMQ. Are you planning on developing a robust embedded message
broker that will be internal to Hookbox?

Interesting nonetheless.

On 26 jul, 07:26, Michael Carter <cartermich...@gmail.com> wrote:
> Greetings,
>
> As some of you may know, I've been working on an alternate
> comet/websocket/web-push server called Hookbox (http://hookbox.org) for the
> past year. The goal of the project is to expose much higher-level apis than
> Orbited, and provide tight integration with existing web technologies, like
> PHP, Django, Rails, Java Servlets, etc.
>
> Please take a look at an introductory article/tutorial that I just published

> on cometdaily:http://cometdaily.com/2010/07/26/a-fast-introduction-to-hookbox/. I hope

[simexous]

Very interesting concept.. How does this compare to Orbited and is it
a viable/stable alternative?
How about scalability and performance?

Thanks..and great job!

--
M.

On Jul 26, 6:26 am, Michael Carter <cartermich...@gmail.com> wrote:
> Greetings,
>
> As some of you may know, I've been working on an alternate
> comet/websocket/web-push server called Hookbox (http://hookbox.org) for the
> past year. The goal of the project is to expose much higher-level apis than
> Orbited, and provide tight integration with existing web technologies, like
> PHP, Django, Rails, Java Servlets, etc.
>
> Please take a look at an introductory article/tutorial that I just published

> on cometdaily:http://cometdaily.com/2010/07/26/a-fast-introduction-to-hookbox/. I hope

[Michael Carter]

Hi Leon,

On Mon, Jul 26, 2010 at 6:00 AM, leon <aeond...@hotmail.com> wrote:

Michael,

I took a look at the documentation and would like to ask whether you
plan on handling the message broker within Hookbox itself. I am
currently using Orbited and on the production server I went with
ActiveMQ. Are you planning on developing a robust embedded message
broker that will be internal to Hookbox?

Hookbox is a message broker itself. In the future we may allow you to replicate to amqp or stomp brokers. At the moment the project is geared applications which can be sharded to about 10k users (or less) per instance.

-Michael Carter

[Michael Carter]

On Mon, Jul 26, 2010 at 10:49 AM, simexous <moh...@amtoun.com> wrote:

Very interesting concept.. How does this compare to Orbited and is it
a viable/stable alternative?
How about scalability and performance?

Hookbox is much higher level than Orbited, and provides far better integration with existing web technologies. Hookbox tends to be far more opinionated than Orbited. You can't use your own message queue, or your own protocol: You do it hookbox's way or choose something else.

Hookbox is built with eventlet which runs on top of epoll/kqueue/etc., so you should have very good performance. The project has been under development for about a year, and a solid set of the core features are ready for production.

-Michael Carter

[leon]

So I dare guessing you're preparing a broker for HTML 5.

Have you considered pluggable functionality, say, in case I want to
use stomp to talk to Hookbox and I don't want to create a separate
layer/server in between? It would be nice if messages could be
internally forwarded to an interpreter and then re-formatted to the
broker's standard.

My other take on this is authentication, but it seems you've got it
covered already.

As for stability, how heavy were the tests performed (mechanical load,
bandwidth, memory leaks and so on)? I like ActiveMQ for its HP (I
guess it runs on diesel) but I am not fond of Java given its weight on
the host system. Also, how much memory does Hookbox use per channel
created?


Let there be Comet, and Push away.
Leon

On Jul 26, 3:05 pm, Michael Carter <cartermich...@gmail.com> wrote:
> Hi Leon,
>

[Onur CELEBI]

Sounds very promising, congratulations Michael. I will try it by integrating instant update in some parts of webpages of an existing django application I did (not spool (which already have orbited+xmpp, another one). I'm wondering how easy it is to integrate.

 

"Hookbox is built with eventlet which runs on top of epoll/kqueue/etc., so you should have very good performance"

There is something I don't understand, the power of orbited's scalability was based on horizontal scalability (of some other tcp servers such as xmpp, as it was only a proxy). Although hookbox have good vertical scalability, what about the horiwontal one? Will we be able to create a cluster or something to decrease the charge of each hookbox?

 

--
You received this message because you are subscribed to the
Orbited discussion group.
To post, send email to
<orbite...@googlegroups.com>
To unsubscribe, send email to
<orbited-user...@googlegroups.com>
For more options, visit
<http://groups.google.com/group/orbited-users>

[Michael Carter]

On Tue, Jul 27, 2010 at 6:14 AM, Onur CELEBI <on.c...@gmail.com> wrote:

 

"Hookbox is built with eventlet which runs on top of epoll/kqueue/etc., so you should have very good performance"

There is something I don't understand, the power of orbited's scalability was based on horizontal scalability (of some other tcp servers such as xmpp, as it was only a proxy). Although hookbox have good vertical scalability, what about the horiwontal one? Will we be able to create a cluster or something to decrease the charge of each hookbox?

The scalability goals of hookbox are basically this: If you need 5,000 concurrent users, then don't worry about it; If you need 50,000 concurrent users, then buy a top-end server and don't worry about; If you need 500,000 concurrent users, then shard your application such that you can keep certain users/groups on specific hookbox instances. 

-Michael Carter

[Onur Celebi]

ok i see, as channels don't have to communicate with other channels, we can deploy more than one hookbox instances, each of them having his own set of channels.

it's really a good idea :)

sent from my super iphone

--