ratio degrades the system performance since this case needs the content transmission from a content sever to a transcoding proxy and the complete transcoding task.
Figure 4 shows the cache hit ratio of each scheme. The cooperation schemes provide much higher ratio of both an exact hit and a useful hit. The hit ratio of the content-aware scheme is slightly higher than the query-based scheme. In the query-based scheme, the exact hit ratio of the local cache is 9.01% and the exact hit ratio of the remote cache is 23.89%. In the content- aware scheme, the exact hit ratio in the private storage is only 6.71%, which is smaller than the query-based, but the exact hit ratio of the public storage is 29.35%, which is much bigger than the query-based. Even if the local exact hit ratio of the content-aware scheme is smaller, the main factor of high exact hit ratio is the remote exact hit ratio for both schemes. In this case, the global lookup process of the query-based scheme causes the long decision time due to two-phase lookup in the multiple-version environment mentioned in the BACKGROUND section. However, the content-aware scheme finds the exact object with the simple discovery process, which takes only one query to the home proxy. Therefore, the content-aware scheme can provide better performance than the query-based scheme.
We can see that the useful hit ratio is increased in case of the content-aware cooperation architecture. The reason is that each useful version is clustered to be discovered directly in the content-aware coopera- tion architecture and they use the cache storage more efficiently without the redundant copies. Additionally, in the content-aware scheme, the local useful hit ratio is zero because the private storage is used to find the exact objects only.
FuturE trEndS
The transcoding proxies help each other and achieve high performance. However, there is room for enhance- ment of the trancoding performance. Some researches (Chang & Chen, 2003; Lum & Lau, 2002) explore the efficient methods such as cache replacement and content adaptation to increase the performance of transcoding in the proxy cache and these methods can be used with the cooperative transcoding proxy
caches. These studies are motivated by the various heterogeneous environment of internet. Moreover, because of a increasing in the ubiquitous computing, the content-aware approaches are more important to support the various devices and users. The middleware to support the ubiquitous computing should concern not only the content-awareness but also the cooperation of the servers to share the data.
concLuSIon
In this article, we propose the efficient cooperative transcoding proxy architecture, which uses the content- aware caching. We cope with the problem that is caused by the multiple version environments by using the content-aware caching, which means that every version for an object is stored at one designated proxy together. The proposed architecture makes the communication protocol between each proxy simpler, especially the discovery process and reduces the number of mes- sages, which are used to maintain the cache system such as ICP queries and object responses. Moreover, because of gathering all versions of an object at one proxy, the exact hit ratio increases and the performance of the system increases too. This architecture has an improvement of 20 percentage points of response time, an increase of 10 percentage points of cache hit ratio, and improved scalability on bandwidth consumption. Though the many advantages exist, the hot spot problem can be appeared. We prevent this problem by using the private cache and the cache update policy.
rEFErEncES
Canali, C., Cardellini, V., Colajanni, M., Lancellotti, R., & Yu, P. S. (2003). Cooperative architectures and algo- rithms for discovery and transcoding of multi-version content. Proceedings of the 8th International Workshop
on Web Content Caching and Distribution.
Cardellini, V., Colajanni, M., Lancellotti, R., & Yu, P. S. (2003). A distributed architecture of edge proxy servers for cooperative transcoding. Proceedings of the 3rd
IEEE Workshop on Internet Applications, San Jose.
Cardellini, V., Yu, P. S., & Huang, Y. W. (2000). Col- laborative proxy system for distributed Web content transcoding. Proceedings of the 9th ACM International
Content-Aware Caching for Cooperative Transcoding Proxies
Conference on Information and Knowledge Manage- ment (pp. 520-527).
Chang, C. & Chen, M. (2003). On exploring aggregate effect for efficient cache replacement in transcoding proxies. IEEE Transactions on Parallel and Distributed
Systems, 14(6), 611-624.
Chankhunthod, A., Danzig, P. B., Neerdaels, C., Schwartz, M. F., & Worrell, K. J. (1996). A hierarchical Internet object cache. Proceedings of the 1996 Usenix
Technical Conference, San Diego.
Howell, F., & McNab, R. (1998). SimJava: A discrete event simulation package for Java with applications in computer systems modeling. Proceedings of the 1st
International Conference on Web-based Modelling and Simulation, San Diego.
IRCache project (2003) http://www.irchache.net Knutsson, B., Lu, H., & Mogul, J. (2002). Architec- tures and pragmatics of server-directed transcoding.
Proceedings of the 7th International Workshop on Web Content Caching and Distribution.
Lindemann, C., & Waldhorst, O. P. (2001). Evaluat- ing cooperative Web caching protocols for emerging network technologies. Proceedings of International
Workshop on Caching, Coherence and Consistency.
Lum, W. Y., & Lau, F. C. M. (2002). On balancing between transcoding overhead and spatial consumption in content adaptation. Proceedings of the 2002 ACM
MOBICOM Conference.
Maheshwari, A., Sharma, A., Ramamritham, K., & Shenoy, P. (2002). TransSquid: Transcoding and caching proxy for heterogeneous e-commerce environments.
Proceedings of the 12th IEEE International Workshop on Research Issues in Data Engineering (pp. 50-59).
Singh, A., Trivedi, A., Ramamritham, K., & Shenoy, P. (2002). PTC: Proxies that transcode and cache in heterogeneous Web client environments. Proceedings
of the 3rd International Conference on Web Information Systems Engineering.
KEY tErMS
Home Proxy: A proxy which is responsible for a
target object key.
Node ID: A unique and hashed value to identify
a node.
Object ID: A unique and hashed value to identify
an object.
Private Storage: A storage on a proxy cache for
the clients which are only on this proxy cache.
Proxy Cache: A server which locates on the border
of ISP or institution and helps the original server by intercepting the user request.
Public Storage: A storage on a proxy cache for the
group of clients which are on the cooperative proxy caches.
Transcoding: Transform the size, quality, presen-
tation style, etc., of objects to meet the capabilities of clients.