Improving Data Access in Mobile and Wide Area Environments
The rapid growth of the Internet has increased both the number and the
diversity of clients who access data on remote servers, in both mobile
and fixed network environments. Remote data access on fixed networks
is characterized by high latency and frequent updates, e.g. stock
quotes. An important characteristic of Internet data access is that
clients often have different requirements for the latency and recency
of objects. However, there has been little work that aims to
accomodate client needs.
Caching can improve client latency but it has the disadvantage that
the cached data becomes stale. Existing approaches to keeping cached
objects up to date include TTL, server side invalidation, and
background prefetching. TTL assigns each object a Time-to-Live and
validates any object whose TTL has expired, which may add unnecessary
latency to a request. Server side invalidation guarantees that data
in the cache is kept up-to-date, but requires cooperation from remote
servers and does not scale well. Finally, prefetching objects in the
background can reduce latency, but may not keep cached objects
sufficiently up-to-date. To summarize, existing approaches do not
consider diverse client preferences and may add unnecessary overhead
in terms of latency or bandwidth, or may deliver data that is less
recent than the client requires.
In the first part of this research we introduce Latency-Recency
Profiles that allow clients to express target values for the desired
latency and recency of objects. We define a scoring function that
uses the target values to determine when to download an object or when
to use a cached copy. The function can be tuned to meet the
target values and can provide guarantees with respect to the maximum
latency or recency of requested objects.
Experiments with Web trace data and synthetic data validate the
effectiveness of our approach and show that the Profile-based
framework can provide substantial bandwidth savings in a web
environment compared to existing approaches, while still providing recent data in many cases.
In the second part of this research we use this Profile-based scheme
to improve data access in mobile environments. In mobile
environments, clients typically connect to a wireless base station to
access data on fixed networks. The base station schedules objects for
delivery to clients via the wireless downlink. Since the requests of
multiple clients are multiplexed through the base station, maintaining
a proxy cache near the base station can leverage commonalities in
client requests and reduce access latencies. We present a complete
framework that incorporates client targets for latency and recency
into the cache utilization, downloading, and scheduling decisions at a
mobile base station. Our results show that compared to
algorithms that do not consider client preferences, our scheme
improves the resource utilization on the wireless downlink as well as
the fixed network access link.
More information is available in the following papers:
Bright L., Raschid L. Using Latency-Recency Profiles for Data Delivery on the Web. Proceedings of the Conference on Very Large Data Bases (VLDB), 2002.
Bright L., Bhattacharjee S., Raschid L. Supporting Diverse Mobile Applications with Client Profiles Proceedings of the ACM Workshop on Wireless Mobile Multimedia (WoWMoM), 2002.
For additional papers, see my publications.