| Abstract |
iSCSI is proposed as a possible solution to building future storage systems. However, using iSCSI raises numerous questions about its implications on system performance. This lack of understanding of system I/O behavior in modern and future systems inhibits providing solutions at the architectural and system levels. First of all in this work, we try to understand the behavior of the application server (iSCSI initiator), to evaluate the overhead introduced by iSCSI compared to systems with directly-attached storage, and to provide insight about how future storage systems may be improved. We examine these questions in the context of commodity iSCSI systems that can benefit most from using iSCSI. Our analysis shows that building next generation, network-based I/O architectures, requires optimizing I/O latency, reducing network and buffer cache related processing in the host CPU, and increasing the sheer network bandwidth to account for consolidation of different types of traffic. Based on this knowledge we provide our own solution for a scalable distributed storage system, Orchestra. Orchestra, is a system for cluster storage virtualization that allows sharing of storage volumes at the block level by providing block locking and free block allocation services. Orchestra is a separate software infrastructure and it is not part of this work. To allow file-based applications access to Orchestra block devices, we design and implement a stateless, pass-through file system, Orchestra-fs, that relies on Orchestra for locking and block-allocation support and allows multiple existing applications to use a single Orchestra volume. We present the implementation of Orchestra-fs under Linux and evaluate it over various setups using both single application and multiple storage nodes. We find that since most functionality is implemented at the block layer, which is well structured, there is little overhead beyond TCP/IP communication costs and existing kernel overheads for disk I/O. Finally, our results show that performance scales as the number of storage nodes grows for experiments where there is no file sharing between applications.
|
Ανάλυση της Απόδοσης και Κλιμάκωσης ενός Δικτυακού Διαμεριζόμενου Αποθηκευτικού Χώρου σε Επίπεδο Block
