25 karma · joined August 9, 2010
Also you should note, that this font was created by a dyslexic, so I think he knows what he is doing.
Given all this, I wasn't able to go beyond 2k simultaneous connections. Reasons? Our Rails setup wasn't able to handle that much requests. For the sake of testing I wrote a synthetic tests where Rails requests were mocked. RabbitMQ was running on Thinkpad Edge (quad core i5@2.6Ghz), node.js client and server was running on dual core pentium@3Ghz. This configuration gave me 3.5-4.5 thousands of message deliveries per second, both PCs were at the top of capacity. Such performance looked suspiciously low for me, so I went further. Reading node-amqp shown that it is totally flawed (i.e. it is allocating 130kb buffer for each message sent). Doing some tweaking I managed to increase message delivery rate to almost 15 thousands of messages/second.
Strange part of the outcome is that processing both 4k and 14k messages created almost identical load on RabbitMQ.
Lessons learnt: 1. RabbitMQ is very robust and easy to work with. 2. Do not measure RabbitMQ's performance by LA or CPU consumption. If you feel stuck, try optimizing something. 3. V8 memory allocation/GC suck. My server was constantly crashing due to std::bad_alloc being thrown after next packet buffer allocation. 4. node-amqp suck even more. I am going to work on it somewhen soon. 5. Node.js on it's own is a toy, not a tool. Do not build mission critical systems designated to handle long living connections with it.
World is complex. Technologies are. Deal with it.
Another thing that seriously bugs me is the approach to mask completely different goals and processes by imaginary simplicity of overlaying technology stack. Writing a web app and making it scale are totally different goals, choose what are you going to do first. Textual documents storage will differ from video hosting by a couple of orders of magnitude. The more generalization you achieve, the less freedom you'll have.
The fact that you've used loosy implementation doesn't mean that the spec itself is a crap.
DLNA is not a silver bullet and has some flaws, but it is thousand times better then AirPlay just because it is more open and does not require specific hardware to accomplish simple tasks. And what is even more important, at the moment huge amount of work is being done worldwide to enrich the list of devices supporting these protocols.
Can you please explain what do you mean?
Anyhow, at the moment opensource is only starting to be curious about something other then writing code and compiling it. Too bad they are doing new stuff kind of wrong (see KDE4 and Gnome3, my two major reasons to use Fluxbox).
And there's more! This December I spent a week in Samsung's R&D working with their extension of UPnP that will deliver custom user interfaces for the TVs. They will be like regular javascript enabled web pages with an access to TV's media playing capabilities. No Apple, still.
So I beg you, Apple fans, try to take a broader look before shouting out loud about new killer features 'invented' by Apple.
Oh, btw Goole Books snippets are getting injected into my search results more and more often, so it looks like the idea of searching through the books isn't dead at all.