28-11-2012, 05:32 PM
Upload Cache in Edge Networks
Upload Cache in Edge Networks.pdf (Size: 293.7 KB / Downloads: 22)
Abstract
Research efforts have been put into content retrieval
in the Internet, ranging from traditional web proxy
to recent content-oriented network architectures. With
the emerging trends of uploading large user-generated
content, we argue that Internet should not only aid end
users in downloading content from steadily available
servers but also facilitate uploading content. In this
paper, we propose upload cache in edge networks, a
new edge-network mechanism assisting upload of usergenerated
content (UGC). Our proposed mechanism
brings benefit for both end users and service providers.
For end users, it shortens the duration while user must
stay online for uploading their generated content. Also
for service providers, it reduces peak traffic volume
between edge networks and data centers by slightly
shifting the upload timing without incurring much extra
latency overhead added. Our analysis with replaying
the captured traffic shows that this mechanism reduces
upload tether time of 24% end users by more than
half and flattens the traffic peak for the access service
provider by 37%.
Introduction
The Internet has been originally designed for the
model of host-to-host data communication. However,
people have observed apparent trends of content domination
in the current Internet usage. Research efforts
have been put into content retrieval in the Internet,
which dates back to the web proxy caches [1] that
have been comprehensively studied since decades ago.
Other popular topics in this field include wide-area
network (WAN) accelerator [2], content distribution
networks (CDN) [3] and so on. Recently, researchers
are exploring the redesign of the Internet communication
model to follow the content domination trend,
and their achievements are generally called contentoriented
network architectures [4], [5], [6].
The common argument for most of these solutions
is that they aid end users within edge networks in
retrieving content from steadily available servers, while
little research has focused on uploading user-generated
content (UGC). Recent studies show that the traffic
uploading UGC accounts for a large portion of the
current Internet traffic and brings new challenges in
the Internet of today [7], [8].
Design
Unlike the traditional content upload schemes where
only a pair of client and server is involved in an
upload procedure, a gateway in the edge network plays
an important role in the proposed mechanism. In our
proposal, an end user uploads its generated content to
the gateway located within the same edge network with
the user. The gateway caches the received content and
uploads it to the corresponding destination server.
There are two major goals of the proposed mechanism.
The first goal is to shorten the duration while an
end user must stay online for uploading their generated
content, making use of the short RTT and ample
bandwidth within the edge networks. The second goal
is to flatten the traffic peak in the network of service
providers including edge networks and the destination
servers.
Scheduled Upload
In most cases, end users do not need the instant
availability of uploaded content. In fact, instant availability
is sometimes impractical to be provided, such as
uploaded video content usually need to be re-encoded
to some specific formats. Therefore, we can schedule
the actual upload from the gateway to the destination
server to a specific time negotiated, rather than let the
gateway upload the content to the destination server
as soon as it has received the complete content, so
that we can avoid creating a large amount of upload
traffic when the edge network or the destination server
already has a high volume of network traffic.
There are two cases to consider regarding the tradeoff
between instant availability and traffic control. Access
service providers may choose to schedule actual
upload to flat the traffic peak for the edge network (that
is, itself) or the destination servers (in which case there
should be some contract so that it also benefits the
access service provider). In the former case, the gateway
asks the client for the deadline indicating that the
client can tolerate the actual upload process begins by
that time, while it accepts the content from the client,
and then schedules the upload to some time before the
deadline according to its traffic prediction.
Server Delegation
In traditional caching systems, edge networks need
to retrieve the content at least once to populate the
cache, which limits the efficiency of download cache in
reducing downstream traffic. In contrast, our proposed
mechanism saves even the first content retrieval by
means of cache reuse.
With the proposed mechanism, after data of the
content is uploaded to the destination server, the server
can reply to the gateway with the publish address of
the content, such as the URI where the content will
be accessible, if the content is cache-able. With the
publish address, the gateway can to respond to the
requests to the cached content in delegation of the destination
server, or reuses the cached upload data for the
download caching system in the same edge network.
As the past research [10] on the geographic distribution
of content request proves spatial locality in content
references, the cache reuse feature is highly effective
in improving the efficiency of reducing downstream
traffic.