Hi Andrzej, OK, good. I think we'll want to reboot the libzmq internals at some point, perhaps based on nanomsg, and that would be a moment to put in better memory management. Or perhaps based on the disruptor pattern. To be seen, it doesn't really matter once we have the protocol worked out.
There are a few things I'd still like to fix in the protocol (like heartbeating), but it needs more pliable stacks. We're seeing quite a few ZMTP stacks springing up, e.g. https://github.com/spotify/netty-zmtp. This is part of the motivation for the June meeting in Brussels: to discuss the technical future of libzmq and where to focus. -Pieter On Mon, May 6, 2013 at 10:45 AM, Andrzej Dworak <[email protected]> wrote: > Hi Pieter, > > I really like your approach. For us, with a few thousand servers distributed > over a few thousand computers, the compatibility between new versions of the > protocol is crucial. For instance, now, we need to provide an > interoperability layer to connect new (ZMQ) and old (CORBA) > services/servers/clients. That's a huge effort. I can't really imagine doing > it again :) So, without protocol versioning we would be simply stuck with a > certain version of ZMQ. > > I also like all the new upgrades, especially the additions concerning the > security. We run on a private network and for internal purpose we have our > own security layer on-top... But it's impossible to address efficiently some > of the problems that the new protocol should help to solve. > > As for the memory allocation, for us embedded systems are as important as > hard RT... and then, the only solution is to set constraints and preallocated > on start-up. But this calls for another big discussion. > > Cheers, > Andrzej > > > ________________________________________ > From: [email protected] > [[email protected]] on behalf of Pieter Hintjens > [[email protected]] > Sent: 02 May 2013 15:56 > To: ZeroMQ development list > Subject: Re: [zeromq-dev] a hot discussion on ZMQ vs YAMI; evaluation of > middleware we did at CERN at the end of 2011 > > Hi Andrzej, > > Thanks for this PDF, it's really interesting. > > I'm curious to know your opinion of the work we're doing on the wire > protocol (ZMTP), which is now in v3.0 in draft stage > (rfc.zeromq.org/spec:23). The goal is to ensure that different > implementations can talk together safely, over a long timeframe. > > -Pieter > > > On Thu, May 2, 2013 at 3:36 PM, Andrzej Dworak <[email protected]> wrote: >> Dear colleagues, >> >> First of all I should have sent that paper more than a year ago, sorry. >> Hopefully it will be still of some value... In fact it was only a very >> interesting discussion at reddit that reminded me to send the small review >> we did. Again, sorry for being late on that. >> >> http://www.reddit.com/r/programming/comments/1d0pwj/yami4_vs_zeromq/ >> >> I hope 150k is not an issue for any box... >> To sum up the paper: we have evaluated a number of middlewares in the >> context of our requirements. A short description of each product is provided >> and how it would fit into our system. The result - ZMQ fits perfectly in our >> environment, better than anything else. For details on the requirements* and >> why ZMQ is better for us than anything else with have tried see the paper. >> >> Best, >> Andrzej >> >> *I guess our reqs are quite common to any big scalable distributed system >> that needs to survive the next +10 years. >> >> _______________________________________________ >> zeromq-dev mailing list >> [email protected] >> http://lists.zeromq.org/mailman/listinfo/zeromq-dev >> > _______________________________________________ > zeromq-dev mailing list > [email protected] > http://lists.zeromq.org/mailman/listinfo/zeromq-dev > _______________________________________________ > zeromq-dev mailing list > [email protected] > http://lists.zeromq.org/mailman/listinfo/zeromq-dev _______________________________________________ zeromq-dev mailing list [email protected] http://lists.zeromq.org/mailman/listinfo/zeromq-dev
