Central processing unit cache friendly multithreaded allocation

Abstract

A cluster allocation bitmap determines which clusters in a band of storage remain unallocated. However, concurrent access to a cluster allocation bitmap can cause CPU stalls as copies of the cluster allocation bitmap in a CPU's level 1 (L1) cache are invalidated by another CPU allocating from the same bitmap. In one embodiment, cluster allocation bitmaps are divided into L1 cache line sized and aligned chunks. Each core of a multicore CPU is directed at random to allocate space out of a chunk. Because the chunks are L1 cache line aligned, the odds of the same portion of the cluster allocation bitmap being loaded into multiple L1 caches by multiple CPU cores is reduced, reducing the odds of an L1 cache invalidation. The number of CPU cores performing allocations on a given cluster allocation bitmap is limited based on the number of chunks with unallocated space that remain.

Description

BACKGROUND[0001]Computer storage needs continue to increase, both in terms of capacity and performance. For many years hard disk drives based on rotating magnetic media dominated the storage market, providing ever increasing density and throughput combined with low latency. However, for certain applications even better performance was desired, and so solid-state drives (SSDs) were introduced that out-performed traditional hard drives, yet cost significantly more per byte of storage.[0002]Some computing applications are more sensitive to differences in storage performance than others. For example, core operating system functions, low latency applications such as video games, storage focused applications such as databases, and the like benefit more from the increased performance of an SSD than web browsing, media consumption, and other less storage intensive tasks. Similarly, computing tasks that perform a significant number of random access storage operations, as opposed to streaming...

AI Technical Summary

Benefits of technology

This patent aims to improve the performance of hybrid storage devices, which combine multiple physical storage devices formatted as a single logical filesystem volume. By selectively storing different types of data on differently performant tiers, the overall performance of the device can be significantly impacted. The patent suggests that metadata, which controls how data is stored and retrieved from a storage device, should be preferentially stored on higher performant tiers to improve overall filesystem performance. The technical effect of this patent is improved performance and efficiency of hybrid storage devices by optimizing metadata operations and selectively storing data on differently performant tiers.

Problems solved by technology

Metadata has a central role in filesystem operations, and metadata operations are often performed while holding locks, increasing filesystem latency.

Images