The non-volatile memory (NVM) which can provide DRAM-like performance and disk-like persistency has the potential to build single-level systems by replacing both DRAM and disk. Keeping data consistency in such systems is non-trivial because memory writes may be reordered by CPU. Although ordered memory writes for achieving data consistency can be implemented using the memory fence and the CPU cache line flush instructions, they introduce a significant overhead (more than 10X slower in performance). In this paper, we focus on an important and common data structure, B$^+$Tree. Based on our quantitative analysis for consistent tree structures, we propose NV-Tree, a consistent, cache-optimized and workload-adaptive B$^+$Tree variant with significantly reduced consistency cost (up to 96 percent reduction in CPU cache line flush). To further optimize NV-Tree under various workloads, we propose a workload-adaptive scheme in which the sizes of individual nodes can be dynamically adjusted to improve the performance over time. We implement and evaluate NV-Tree and NV-Store, a key-value store based on NV-Tree, on an NVDIMM server. NV-Tree outperforms the state-of-art consistent tree structures by up to 12X under write-intensive workloads. NV-Store increases the throughput by up to 7.3X under YCSB workloads compared to Redis.

With more applications moving to cloud, scalable storage systems, composed of a cluster of storage servers and gateways, are deployed as the back-end infrastructure to accommodate high-volume data. In such an environment, it is a challenge to provide predictable and controllable storage performance for multitenanted users with multiple applications, due to performance violation from misbehaving applications. In this paper, we propose a two-level QoS controller over scalable storage system. On the higher level, I/O throughput rented by each tenant is guaranteed and strictly limited by a CAP value. On the lower level, this rented service can be on-demand served among multiple applications under the same tenant. Thus our distributed controller not only shields performance violation from "noisy" tenants but also allows tenants to fully utilizing the rented I/O throughput. Furthermore, the QoS controller is implemented in an efficient manner, by reusing the communication channels among gateways and storage servers and piggybacking control signals on data communications. The experimental results have shown that the two-level QoS controller can guarantee I/O throughput at tenant level by controlling the CAP value while accelerating applications by on-demand serving at a very little computation cost.

Efficient and scalable distributed metadata management is critically important to overall system performance in large-scale distributed file systems, especially in the EB-scale era. Hash-based mapping and subtree partitioning are state-of-the-art distributed metadata management schemes. Hash-based mapping evenly distributes workload among metadata servers, but it eliminates all hierarchical locality of metadata. Subtree partitioning does not uniformly distribute workload among metadata servers, and metadata needs to be migrated to keep the load balanced roughly. Distributed metadata management is relatively difficult since it has to guarantee metadata consistency. Meanwhile, scaling metadata performance is more complicated than scaling raw I/O performance. The complexity further rises with distributed metadata. It results in a primary goal that is to improve metadata management scalability while paying attention to metadata consistency. In this paper, we present a ring-based metadata management mechanism named Dynamic Ring Online Partitioning (DROP). It can preserve metadata locality using locality-preserving hashing, keep metadata consistency, as well as dynamically distribute metadata among metadata server cluster to keep load balancing. By conducting performance evaluation through extensive trace-driven simulations and a prototype implementation, experimental results demonstrate the efficiency and scalability of DROP.

Efficient and scalable distributed metadata management is critically important to overall system performance in large-scale distributed storage systems, especially in the EB era. Traditional state-of-the-art distributed metadata management schemes include hash-based mapping and subtree partitioning. The former evenly distributes workload among metadata servers, but it eliminates all hierarchical locality of metadata. It cannot efficiently handle some operations, e.g., renaming or moving a directory that requires metadata to be migrated among metadata servers. The latter does not uniformly distribute workload among metadata servers, and metadata need to be migrated to keep the load balanced roughly. In this paper, we present a ring-based metadata management scheme, called Dynamic Ring Online Partitioning (DROP). It can preserve metadata locality using locality-preserving hashing, as well as dynamically distribute metadata among metadata server cluster to keep load balancing. By conducting performance evaluation, experimental results demonstrate the effectiveness and scalability of DROP.

In this paper, we consider the design of a dynamic storage resource management framework for the grid. The framework is service-oriented and conforms to the Open Grid Service Infrastructure (OGSI) specifications. The higher-level framework design is inspired by the grid and the peer-to-peer network architecture, which have been proven to be very effective in distributed environments. The design guarantees a robust way of maintaining the storage resources as it does not rely on any supporting infrastructure. In our design, we included QoS performance metrics as part of our service interface to cater for high-performance grid applications. At the lower level, we consider the usage of Storage Management Initiative Specifications (SMI-S), which enhance the flexibility and management of heterogeneous storage systems/devices, of various vendors.