[Orbited] Re: Our experiences so far with Orbited..

Could you please explain how you handled security from the ajax calls
to the message broker?

On Dec 14 2009, 10:09 am, BobDobbs <[EMAIL PROTECTED]> wrote:
> I am currently checking with my managers about what I could reveal. We
> are a full blown commercial operation.. and some of our clients are
> fairly high profile. At this time... I am not sure what I could say
> since some of our clients might not want their new systems getting
> attention just yet (or their link to our company). It is not that our
> company is secretive or closed-minded... it is just that I am not sure
> what my bounds are at the moment.
>
> I will get some more info.. and at the very least... I can give you a
> decent amount of information offline. For now I can say that we ported
> a legacy Win32 application to the web. The purpose is real-time
> bidding. We have three main components... the bid server.. the
> statemachine.. and the web UI.
>
> The web UI speaks strictly to the statemachine using Orbited pointing
> at ActiveMQ. The statemachine is a layer between us and the legacy
> server (using ActiveMQ).
>
> We have fairly heavy messaging. Messages are coming down to the UI
> fairly rapidly (much like an auction if you have heard a live one). We
> really could trim the messaging down a bit... but since Orbited is a
> champ... we haven't felt the pressure to do so yet. Also.. upstream
> messages can have many types... but the main two are bids and
> heartbeats. Bids are pretty spaced out (not many items actually have
> 50 bidders all fighting for it). Heartbeats however are heavy. We run
> maybe 100 auctions a day and for each auction... every 20 seconds the
> UIs for a lane will get a heartbeat request. They will then all send a
> single heartbeat back up. This is to make sure that clients are
> connected. These waves of heartbeats coming up gave us a bit of
> trouble. However... Orbited had nothing to do with it and didn't even
> seem to break a sweat from the waves of messages.
>
> Our goal about a year ago was to port this legacy app and have it
> perform exactly the same. As I am sure many in this group are aware...
> its not so easy to get full blown application features into a web
> page. One big restriction is that it still had to keep sub-second
> latency. Magically.. and due to the Orbited team's hard work... we
> have met that goal.. even with thousands of users.
>
> Anyway... that's all I can really say now but I will get some
> direction from my managers and let you know more if I can.
>
> On Dec 12, 3:02 am, Mario Balibrera <[EMAIL PROTECTED]> wrote:
>
>
>
> > Hey Bob. Great to hear Orbited is working out for you. Are you at liberty to
> > tell us more about your project? Would your team mind being listed 
> > here:http://orbited.org/wiki/Sites?Thanks for sharing your experiences with 
> > the
> > community.
>
> > -mario
>
> > On Fri, Dec 11, 2009 at 11:47 AM, BobDobbs <[EMAIL PROTECTED]> wrote:
> > > Well... its a little complicated. We have a large real-time system
> > > that integrates with many other components. I believe that a "node" is
> > > a group of 6 or 7 servers... only one of which has Orbited on it. I
> > > think these servers are pretty strong... dual quad core... probably
> > > like 8 gigs of memory. The box with Orbited also runs a JBoss server
> > > and ActiveMQ and probably not much else. I am pretty sure that even in
> > > the middle of the day... we see 1-5 percent CPU usage in general. I
> > > think the machine might be a little too beefy to really show the
> > > "weight" of Orbited in top. I don't normally look at Orbited's mem/cpu
> > > consumption as we are often more concerned about our java server
> > > component taking memory (it gets to 1.5 gigs at times.. but still only
> > > 3% CPU)
>
> > > In fact... Orbited used to be one of our biggest concerns. However,
> > > since we moved to 0.7.10... we essentially forgot about it because it
> > > never breaks down, slows down, complains or anything. Works like a
> > > charm for thousands of users a day (at least the way we use it)
>
> > > On Dec 11, 2:31 am, Niklas B <[EMAIL PROTECTED]> wrote:
> > > > Hi Bob,
>
> > > > Great to hear your success story, specially for myself that
> > > > (hopefully) a couple of month from rolling on a major Orbited
> > > > application from scratch replacing our old legacy perl apps.
>
> > > > What is your ram usage/cpu for orbited? just curious.
>
> > > > Regards,
> > > > Niklas
>
> > > > On Dec 10, 9:31 pm, BobDobbs <[EMAIL PROTECTED]> wrote:
>
> > > > > Just wanted to give a thanks to the Orbited team.
>
> > > > > We have been moving our legacy client to the web over the past year.
> > > > > We were always a little nervous about Orbited since it is the life
> > > > > line from browser to server.
>
> > > > > We have been in production for about three months now. Every day we
> > > > > have thousands of users on the system sending pretty heavy message
> > > > > loads. Our system requires low latency for all messages. Works like a
> > > > > charm. Roundtrips in Chrome and FF are about 80-160ms on fair
> > > > > connections... IE even provides sub half second latency.
>
> > > > > Most of the time the daemon runs for a month before a restart... and
> > > > > we restart not because of an error or problem.. but because it seems
> > > > > prudent.
>
> > > > > Orbited has met and exceeded any of our expectations so excellent work
> > > > > mi amigos.
>
> > > --
> > > You received this message because you are subscribed to the
> > > Orbited discussion group.
> > > To post, send email to
> > >    <[EMAIL PROTECTED]>
> > > To unsubscribe, send email to
> > >    <[EMAIL PROTECTED]<[EMAIL PROTECTED] glegroups.com>
>
> > > For more options, visit
> > >    <http://groups.google.com/group/orbited-users>

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