Designing Large-Scale Real-Time Data Distribution Networks with ØMQ

Introduction

  • Real-time data feeds?
  • Stock trading
  • TV/radio
  • Weather Reports
  • Social Networks
  • Basically whole state of the world can be turned into real-time feed
  • We often use pull model (req/rep) where push (pub/sub) would be more appropriate because we lack the technology to do so.
  • We still think of computers and networks as glorified hard-disks (serving historical data instead of real-time data).
  • With getting close to 24/7 access to the net the world is changing

What's wrong with real-time feeds?

  • We don't want to saturate the network by a large amount of data that nobody needs
  • We don't care much about load on publishers themselves though
  • The bandwidth problem. How can we eliminate messages nobody is interested in from the network?

Scaling

  • Using devices in the middle

End-to-end nature of filtering

  • Allows for HW intergration
  • Allows for extensible matching algorithms

Conclusion

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License