1 / 19

Dissemination of Dynamic Data on the Internet

Dissemination of Dynamic Data on the Internet. Krithi Ramamritham Pavan Deolasee Amol Katkar Ankur Panchbudhe Prashant Shenoy. Dec 2000. Overview of Presentation. Dynamic Data Temporal Coherency Cache Consistency Push and Pull Approaches Combining Push and Pull

Download Presentation

Dissemination of Dynamic Data on the Internet

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Dissemination of Dynamic Data on the Internet Krithi Ramamritham Pavan Deolasee Amol Katkar Ankur Panchbudhe Prashant Shenoy Dec 2000

  2. Overview of Presentation. • Dynamic Data • Temporal Coherency • Cache Consistency • Push and Pull Approaches • Combining Push and Pull • Dynamically choosing Push or Pull • Conclusions

  3. Dynamic Data Dissemination • Dynamic Data Data which changes rapidly & unpredictably. e.g Sports scores, traffic or weather data. • Coherency requirements Nature of the item User tolerances

  4. How dynamic data is dealt with? • Functions of a Proxy Exploits user specified coherency. Maintains desired temporal coherency. Server Proxy User Push Push Pull

  5. Temporal Coherency • Each cached item must be periodically refreshed. • For highly dynamic data Difficult to maintain cache consistency. Heavy network overload. Server load.

  6. How temporal coherency is achieved. • User is not interested in every change happening at the source. • User specifies a temporal coherency requirement c for each cached item. • Proxy depending on value of c can use push or pull approach to maintain coherency.

  7. Cache Consistency. • Fidelity of the data Degree to which coherency needs of a user are met. • Normally, the problem of cache consistency is resolved by 2 approaches Client-driven Server-driven

  8. Cache Consistency • Client-driven Polling each time Adaptive TTR(time to refresh) • Server-driven Invalidates cache entries Updates proxy cache • In case of dynamic data, Cannot deliver fidelity with optimum resource utilization

  9. The Pull Approach • Each data item is assigned a certain TTR & until that time all requests are satisfied from the cache. • Proxy issues a get request to the server. Periodic Server Push Proxy User Periodic Pull Server Pull Proxy Push User Aperiodic

  10. Pull Approach. • Periodic Pull Proxy periodically polls the server Obtains data with a high frequency • Disadvantages Very high network overhead User may miss some changes Not suitable if rate of change is varying

  11. Pull Approach(Contd..) • Aperiodic pull TTR decreases dynamically when a data item starts changing rapidly Increases when a hot file becomes cold • Adaptive TTR takes into account Rapid changes that have occurred so far Recent changes to the polled data

  12. Push Approach • Server can push data either periodically or aperiodically • Periodic Push Disseminates data based on demand for data item All data items get divided into frequency bands • Disadvantages Low fidelity, Wastage of bandwidth

  13. Push Approach(Contd..) • Aperiodic Push Proxy registers with the server Server uses tcr to determine if data item is to be pushed. Server maintains state information as list of proxies,tcr and last update to each proxy Combines requests with identical tcr intoa single request • Disadvantages Limits scalability, not resilient to failures.

  14. Push v/s Pull • Communication Overhead Pull- Larger load on the network Push-Large message overhead • Computational Overhead High load due to too much polling or monitoring • Space Overhead Maintains c value, pushed value, state with open connection

  15. Push v/s Pull • Resiliency State of server is lost in case of server failures In client failures, resources assigned must be reclaimed • Scalability With upper bound on the number of sockets and state space available, servers become scalable It arises due to excessive server computation and resources allocated

  16. Combining Push and Pull • Leases Contracts given to a lease holder over some property Server informs the client about any changes during the lease period • Adaptive Leases Dynamically adjust the lease duration Gives strong cache consistency

  17. Combining Push and Pull • Server tries to predict when a client is going to poll next • If it knows that a client is going to miss some change it pushes the data to the client • In this approach the performance in terms of fidelity and resiliency can be controlled

  18. Dynamically choosing Push v/s Pull • If resources are plentiful, every client is given a push connection • As clients increase, some clients are shifted to pull mode and scalability is ensured • Clients are assigned priorities by Temporal coherency requirement Fidelity requirement Network bandwidth available

  19. Conclusions • It is a priori difficult to determine whether a push or pull based approach is to be employed. • Combination of Push and Pull is used. • Currently work is going on in determining range of applicability of new algorithms for disseminating web data

More Related