Sunday, February 20, 2011

Volley: Automated Data Placement for Geo-Distributed Cloud Services

Link to the paper: 
http://research.microsoft.com/en-us/people/sagarwal/nsdi2010.pdf

Presenter: Kavyashree Ramakrishna Prasad  
Reviewers: Sandeep Venkatesh and Luquan Huang

4 comments:

  1. In section 3.1, the author states "If the source
    or destination of a request is movable (i.e., because it is a data item under the control of the cloud service), we log a GUID identifier rather than its IP address" How does Volley determine if the source or destination is movable or an end point? What mechanism does it employ?

    ReplyDelete
  2. Since this paper focused on improving latency and not bandwidth, has any current Volley efforts been made to improve bandwidth between arbitrary points in the internet?

    ReplyDelete
  3. Could you please comment on oscillation damping and Volley's take on that which is mentioned in the section 3.3 as I did not quite understand it?

    ReplyDelete
  4. In the paper authors mention that the volley takes 4 additional inputs from the data centre managers while making decisions about placement of the data. But in the Volley's algorithm description there is no mention of these parameters. Can you comment on the phase of the volley's algorithm in which these parameters will be used ?

    ReplyDelete