Performing initiative data prefetching
- 1. PERFORMING INITIATIVE DATA PREFETCHING IN DISTRIBUTED FILE SYSTEMS FOR CLOUD COMPUTING ABSTRACT This paper presents an initiative data prefetching scheme on the storage servers in distributed file systems for cloud computing. In this prefetching technique, the client machines are not substantially involved in the process of data prefetching, but the storage servers can directly prefetch the data after analyzing the history of disk I/O access events, and then send the prefetched data to the relevant client machines proactively. To put this technique to work, the information about client nodes is piggybacked onto the real client I/O requests, and then forwarded to the relevant storage server. Next, two prediction algorithms have been proposed to forecast future block access operations for directing what data should be fetched on storage servers in advance. Finally, the prefetched data can be pushed to the relevant client machine from the storage server. Through a series of evaluation experiments with a collection of application benchmarks, we have demonstrated that our presented initiative prefetching technique can benefit distributed file systems for cloud environments to achieve better I/O performance. In particular, configuration-limited client machines in the cloud are not responsible for predicting I/O access operations, which can definitely contribute to preferable system performance on them. Proposed system: The proposed mechanism first analyzes disk I/O tracks to predict the future disk I/O access so that the storage servers can fetch data in advance, and then forward the prefetched data to relevant client file systems for future potential usages. In short, this paper makes the following two contributions: 1) Chaotic time series prediction and linear regression prediction to forecast disk I/O access. We have modeled the disk I/O access operations, and classified them into two kinds of access patterns, i.e. the random access pattern and the sequential access pattern. Therefore, in order to predict the future I/O access that belongs to the different access patterns as accurately as possible (note that the future I/O access indicates what data will be requested in the near future), two
- 2. prediction algorithms including the chaotic time series prediction algorithm and the linear regression prediction algorithm have been proposed respectively. 2) Initiative data prefetching on storage servers. Without any intervention from client file systems except for piggybacking their information onto relevant I/O requests to the storage servers. The storage servers are supposed to log disk I/O access and classify access patterns after modeling disk I/O events. Next, by properly using two proposed prediction algorithms, the storage servers can predict the future disk I/O access to guide prefetching data. Finally, the storage servers proactively forward the prefetched data to the relevant client filesystems for satisfying future application’s requests. MODULES ASSUMPTIONS IN APPLICATION CONTEXTS PIGGYBACKING CLIENT INFORMATION I/O ACCESS PREDICTION INITIATIVE DATA PREFETCHING ASSUMPTIONS IN APPLICATION CONTEXTS This newly presented prefetching mechanism cannot work well for all workloads in the real world, and its target application contexts must meet two assumptions: Assumption 1: Resource-limited client machines. This newly proposed prefetching mechanism can be used primarily for the clouds that have many resource limited client machines, not for generic cloud environments. This is a reasonable assumption given that mobile cloud computing, which employs powerful cloud infrastructures to offer computing and storage services on demand, for alleviating resource utilization in mobile devices . Assumption 2: On-Line Transaction Processing (OLTP) applications. It is true that all prefetching schemes in distributed file systems make sense for a limited number of read-intensive applications such as database-related OLTP and server-like applications.
- 3. PIGGYBACKING CLIENT INFORMATION Most of the I/O tracing approaches proposed by other researchers focus on the logical I/O access events occurred on the client file systems, which might be useful for affirming application’s I/O access patterns. Nevertheless, without relevant information about physical I/O access, it is difficult to build the connection between the applications and the distributed file system for improving the I/O performance to a great extent. In this newly presented initiative prefetching approach, the data is prefetched by storage servers after analyzing disk I/O traces, and the data is then proactively pushed to the relevant client file system for satisfying potential application’s requests. Thus, for the storage servers, it is necessary to understand the information about client file systems and applications. To this end, we leverage a piggybacking mechanism, which is illustrated in Figure 1, to transfer related information from the client node to storage servers for contributing to modeling disk I/O access patterns and forwarding the prefetched data. I/O ACCESS PREDICTION Many heuristic algorithms have been proposed to shepherd distributing file data on disk storage; as a result, data stripes that are expected to be used together will be located close to one another. An automatic time series modeling and prediction framework to direct prefetching adaptively, this work employs ARIMA (Autoregressive Integrated Moving Average model) time series models to forecast the future I/O access by resorting to temporal patterns of I/O requests. However, none of the mentioned techniques aims to guide predicting physical access via analyzing disk I/O traces, since two neighbor I/O operations in the raw disk trace might not have any dependency. This section will illustrate the specifications about how to predict disk I/O operations, including modeling disk I/Os and two prediction algorithms adopted by our newly presented initiative prefetching mechanism.
- 4. INITIATIVE DATA PREFETCHING The scheme of initiative data prefetching is a novel idea presented in this paper, and the architecture of this scheme is demonstrated, while it handles read requests (the assumed synopsis of a read operation is read (int fields, size t size, off t off)).the storage server can predict the future read operation by analyzing the history of disk I/Os, so that it can directly issue a physical read request to fetch data in advance. The most attractive idea in the figure is that the perfected data will be forwarded to the relevant client file system proactively, but the client file system is not involved in both prediction and prefetching procedures. Finally, the client file system can respond to the hit read request sent by the application with the buffered data. As a result, the read latency on the application side can be reduced significantly. Furthermore, the client machine, which might have limited computing power and energy supply, can focus on its own work rather than predict in related tasks.