Аннотация
Recently, new services have emerged which utilize the
Internet as a delivery mechanism for multimedia content.
With the advent of broadband accesses, more users are willing
to download large volume content from servers, such as
video files of TV shows. While some popular video services
(e.g. YouTube.com) or some broadcasting companies (e.g.
ABC.com) use streaming data with Flash technology, some
media distributors (e.g. iTunes) offer entire TV shows for
download. In this study, we investigate the performance of
the German site OnlineTVRecorder.com (OTR), which acts
as an online video cassette recorder (VCR) where users can
program their favorite shows over a web interface and download
the recorded files from a server or its mirrors. These
files are offered in different file formats and can consist of
several hundred megabytes up to 1GB or more depending on
the length of the TV show as well as the encoding format.
OTR can, thus, be seen as an example for a server-based
content distribution system with large data files.
However, as these server farms are often overloaded, new
requests are queued when the provided download slots are
full. The restriction to a maximum number of simultaneous
downloads guarantees a minimal download bandwidth for
each user. Additionally, the service offers premium users prioritized
access to downloading. The download duration itself
depends on the total capacity of the server and the number
of users currently sharing this capacity. On the other
hand, users who might encounter slow downloads may abort
their downloading attempt if their patience is exceeded.
In this paper, we discuss analytical modeling approaches
which consider the impact of the user’s impatience on the
performance of an OTR server with different file size distributions.
The paper is organized as follows. After describing
the problem and related work, we formulate simple
analytical models and compare their performance in terms
of download duration and success ratio. Especially, we address
the question of how to properly dimension the number
of simultaneous downloads at a server in order to optimize
the performance of the system and to maximize the user
satisfaction.
Пользователи данного ресурса
Пожалуйста,
войдите в систему, чтобы принять участие в дискуссии (добавить собственные рецензию, или комментарий)