Research On The Trustworthy Storage Metering Model For Public Cloud

Cloud platform provides computing and storage resources for users to run tasks, providers charge users for the resources usage to make profits. Storage resource has become more important as the demands of large scale data grow rapidly. However, there is no trustworthy guarantee existed today for both users and providers to justify the exact costs of storage usage. As a result, providers can not prove the correctness of charging, they can also not get the accurate information about the storage usage of the whole platform to help them improve the service. Users can not get accurate charging record to evaluate the performance and verify the amount of resources they actually consume. Moreover, the uncorrect and untrusted charging environment introduces new chances for malicious attackers to achieve resource attacks for cost reducing, which may bring unexpected performance impact to other users. Therefore, discussing how to make an accurate and reliable measurement for the usage of user storages has the important practical significance.After studying the Linux kernel and Xen virtualization architecture, this paper designs and implements a trusted storage measurement modelbased on Xen virtualization architecture. The main points are as follows:(1) Expounds the background, the significance and foreign research status of the topic, and then analyzes the practical significance and the technical challenges to solve this problem.(2) Introduces the architecture of the Linux kernel, and emphatically analyzes the I/O protocol stack and Linux Xen virtualization architecture.(3) Introduces the design and implementation of the trusted storage measurement module.The system uses a trusted storage drive domain to contain the main part of the storage metering work, which sets up information flows in order to get accurate waiting time and execution time for each I/O request,and then adds these information to a log file. The system is build to guarantee the records not to be tampered with, in the meantime, we further make records for users to read, so as to provide them with validation of guarantee.(4) Analyzes the experiment results from two aspects( function and performance). Experimental results show that through the cooperative work of the above modules, the system can correctly reach to the goal and implement the fine-grained reliable measurement for the usage of user storages.(5) Summarizes the work described in this paper, analyzes the advantages and disadvantages in the design, and put forward the improvement direction in the future.