No abstract

Non-Volatile Main Memory (NVMM) devices have been integrated into general-purpose operating systems through familiar file-based interfaces, providing efficient bytegranularity access by bypassing page caches. To leverage the unique advantages of these high-performance media, the storage stack is migrating from the kernel into user-space. However, application performance remains fundamentally limited unless network stacks explicitly integrate these new storage media and follow the migration of storage stacks into user-space. Moreover, we argue that the storage and the network stacks must be considered together when being designed for NVMM. This requires a thoroughly new network stack design, including low-level buffer management and APIs. We propose PASTE, a new network programming interface for NVMM. It supports familiar abstractionsincluding busy-polling, blocking, protection, and run-tocompletion-with standard network protocols such as TCP and UDP. By operating directly on NVMM, it can be closely integrated with the persistence layer of applications. Once data is DMA'ed from a network interface card to host memory (NVMM), it never needs to be copied again-even for persistence. We demonstrate the general applicability of PASTE by implementing two popular persistent data structures: a write-ahead log and a B+ tree. We further apply PASTE to three applications: Redis, a popular persistent key-value store, pKVS, our HTTP-based key value store and the logging component of a software switch, demonstrating that PASTE not only accelerates networked storage but also enables conventional networking functions to support new features.

No abstract

File-system snapshots have been a key component of enterprise storage management since their inception. Creating and managing them efficiently, while maintaining flexibility and low overhead, has been a constant struggle. Although the current state-of-the-art mechanism—hierarchical reference counting—performs reasonably well for traditional small-file workloads, these workloads are increasingly vanishing from the enterprise data center, replaced instead with virtual machine and database workloads. These workloads center around a few very large files, violating the assumptions that allow hierarchical reference counting to operate efficiently. To better cope with these workloads, we introduce Generational Chain Trees (GCTrees), a novel method of space management that uses concepts of block lineage across snapshots rather than explicit reference counting. As a proof of concept, we create a prototype file system—gcext4, a modified version of ext4 that uses GCTrees as a basis for snapshots and copy-on-write. In evaluating this prototype empirically, we find that although they have a somewhat higher overhead for traditional workloads, GCTrees have dramatically lower overhead than hierarchical reference counting for large-file workloads, improving by a factor of 34 or more in some cases. Furthermore, gcext4 performs comparably to ext4 across all workloads, showing that GCTrees impose minor cost for their benefits.

No abstract