An Adaptive And Collaborative Storage System For Virtual Disk Images

In virtualized data centers, virtual disk images(VDIs) serve as the containers of the virtual environment, so their access performance is critical for the overall system performance. Some distributed VDI chunk storage systems have been proposed in order to alleviate the I/O bottleneck of VM management. As the system scales up to a large number of running VMs, however, the overall network traffic would become unbalanced with hot spots on some VMs inevitably, leading to the I/O performance degradation when accessing the VMs.An adaptive and collaborative VDI storage system to resolve the above performance issue is able to dynamically balance the traffic workloads of accessing VDI chunks, based on the run-time network state. Specifically, the system leverages the three-layer architecture based on the two-level cache in local disks of compute nodes. The VDIs split in chunks and distribute across nodes which can be accessed by the VMs in all nodes. Compute nodes with lightly loaded traffic will be adaptively assigned more chunk access requests from remote VMs and vice versa, which can effectively eliminates the above interference problem and thus improves the I/O performance of VMs.We implement a prototype based on our design, and evaluate it by running a well-known benchmark on a real cluster environment. Experiments show that under different network traffic patterns of data centers, our solution achieves up at most 2.6 times performance gain on VM booting time and 8 times on VM's I/O throughput, in comparison to the other state-of-the-art approaches.