1. 15-213 Research Overview TalkOdyssey - 1 Odyssey Agile, Application-Aware Adaptation for Mobility Kip Walker some slides “borrowed” from Satya

2. 15-213 Research Overview TalkOdyssey - 2 A Glimpse of the Future Imagine you are a tourist in Paris  with a wearable computer  wireless access to remote services  unobtrusive heads-up display, microphone, earphones  speech for computer interactions  online language translation Let’s go......

3. 15-213 Research Overview TalkOdyssey - 3 What Makes This Science Fiction? Lack of hardware? NO! We have what we need. Lack of applications? Nope - we have those too. Need a system capable of coping with the problems of mobility Odyssey to the rescue...

4. 15-213 Research Overview TalkOdyssey - 4 Problems with Mobility Mobile elements are resource-poor  relative to static elements of same era  weight, power, size constraints Mobility leads to communication uncertainty  enormous variation in bandwidth & latency  intermittent connectivity Power management is a concern  actions may have to be slowed or deferred  communication costs energy need to rely on resources of remote servers, but may not be able to reach them!

5. 15-213 Research Overview TalkOdyssey - 5 Adaptation Make mobile clients more robust by offering adaptation  rely on servers when possible  function autonomously if needed  monitor and adjust to current conditions

6. 15-213 Research Overview TalkOdyssey - 6 Adaptive Applications applications consume resources network bandwidth, CPU cycles, battery power, disk space, $$$ resources are variable …so… applications adapt use of resources as resource quality changes

7. 15-213 Research Overview TalkOdyssey - 7 Goals Support variety of applications and data types Concurrent applications Quick adaptation Simple programming model

8. 15-213 Research Overview TalkOdyssey - 8 Who Controls Adaptation The Operating System? Individual applications? Both! … Application-Aware Adaptation

9. 15-213 Research Overview TalkOdyssey - 9 What Knobs Do We Have? Where work gets done let powerful remote servers do the work How snazzy the data is: “Fidelity” degrade data meaningfully before giving to mobile host has many dimensions  one is universal: consistency  others depend on data type movies: frame rate, frame quality geographical databases: feature set, minimum feature size  tradeoffs are application-dependent

10. 15-213 Research Overview TalkOdyssey - 10 Cutting to the chase… We built a prototype runs on several UN*X platforms logically an OS extension provides a small API to applications Implementation follows directly from the high-level design need data type aware components to offer fidelity choices need a central piece to watch the resources (network, etc.)

11. 15-213 Research Overview TalkOdyssey - 11 Viceroy and Wardens System-level data differentiation through wardens  specialized code components (a la device drivers)  provides system-level support to manage a data type  trusted entities (unlike applications) Wardens subordinate to viceroy  single, central component  type-independent, system-level support  responsible for all resource allocation, arbitration  central point of authority and control for Odyssey

12. 15-213 Research Overview TalkOdyssey - 12 Client Structure Interceptor Application Odyssey NetBSD OS Kernel Warden2 Warden3 Viceroy Warden1 Odyssey calls Upcall All system calls

13. 15-213 Research Overview TalkOdyssey - 13 Resource Negotiation Applications give viceroy a window of tolerance for some resource  viceroy monitors resource availability  if it leaves window, notifies application via upcall Our architecture supports many resources  we currently focus only on network bandwidth Available bandwidth Fid. 1Fid. 2Fid. 3Fid. 4

14. 15-213 Research Overview TalkOdyssey - 14 Applications Video server offers movie at several levels of fidelity application plays the track that the current bandwidth can support xanim: split into client and server WWW “distillation server” degrades data before shipping to client images can be compressed HTML can be summarized Netscape: client-side proxy + remote distillation server Speech Recognition local/remote/hybrid execution Janus: support remote recognition method, hybrid

15. 15-213 Research Overview TalkOdyssey - 15 Evaluation (don’t blink…) Application-aware adaptation is superior to static strategies applications are able to attain desired “performance”  movie doesn’t drop frames  web delays are masked by compression  speech recognition always available Centralized resource management outperforms alternatives all applications come closer to meeting performance goals Agility needs improvement

16. 15-213 Research Overview TalkOdyssey - 16 Future Work Short term  adaptation for Web objects other than images  improving agility on bandwidth drops  support for unified cache managment Medium term  explore integration of Odyssey in other operating systems  broaden number of managed resources  enlarge range of supported applications ... Long term  deploy Odyssey for real use  dynamic function vs. data shipping as in speech ...

17. 15-213 Research Overview TalkOdyssey - 17 Conclusion Need for adaptation in mobile systems is widely recognized Application-aware adaptation  offers most general and effective approach to adaptation  collaborative partnership between system and application  previous approaches are limiting cases of this approach Odyssey prototype provides initial validation of concept

18. 15-213 Research Overview TalkOdyssey - 18 Contributors to Odyssey Primary contributors  Jason Flinn  Dushyanth Narayanan  Brian Noble  M. Satyanarayanan  Eric Tilton  Kip Walker Numerous secondary contributors involved in  Coda  Janus  NetBSD  Trace Modulation  etc., etc., etc.