CleanCache Merged Into The Linux Kernel
While the first Phoronix benchmarks of EXT4, Btrfs, and XFS on the Linux 2.6.39 kernel were just published this morning, an interesting change was just made for the next Linux kernel that will affect many of the file-systems living within the kernel. For what will be the Linux 2.6.40 kernel, or rather the Linux 3.0 kernel is the finally-merged support for CleanCache.
The CleanCache patch-set began as part of the Transcedent Memory patch-set more than one year ago (and discussed long before that), but with Linux 2.6.40/3.0 it's finally being merged. In fact, the patches just hit the Linus Torvalds Git tree in the past couple of hours. The documentation describes this CleanCache feature as:
There's also this documentation file in the kernel for those wishing to learn more of the technical and implementation details. For the rest, CleanCache simply is designed to offer faster file-system performance. In particular, when there is high memory pressure there should be sizable improvements and for most workloads. "Basically cleancache replaces I/O with memory-copy-CPU-overhead; on older single-core systems with slow memory-copy speeds, cleancache has little value, but in newer multicore machines, especially consolidated/virtualized machines, it has great value."
As part of the push into the linux-2.6.git tree today, there is CleanCache support for the EXT3, EXT4, Btrfs, and OCFS2 file-systems. Other file-systems will also likely be supported in the future. Within the Xen area is also a CleanCache shim to the Transcendent Memory.
It should come as no surprise, but Linux CleanCache benchmarks will be published on Phoronix quite soon.
The CleanCache patch-set began as part of the Transcedent Memory patch-set more than one year ago (and discussed long before that), but with Linux 2.6.40/3.0 it's finally being merged. In fact, the patches just hit the Linus Torvalds Git tree in the past couple of hours. The documentation describes this CleanCache feature as:
This patchset introduces cleancache, an optional new feature exposed by the VFS layer that potentially dramatically increases page cache effectiveness for many workloads in many environments at a negligible cost. It does this by providing an interface to transcendent memory, which is memory/storage that is not otherwise visible to and/or directly addressable by the kernel.
Instead of being discarded, hooks in the reclaim code "put" clean pages to cleancache. Filesystems that "opt-in" may "get" pages from cleancache that were previously put, but pages in cleancache are "ephemeral", meaning they may disappear at any time. And the size of cleancache is entirely dynamic and unknowable to the kernel.
There's also this documentation file in the kernel for those wishing to learn more of the technical and implementation details. For the rest, CleanCache simply is designed to offer faster file-system performance. In particular, when there is high memory pressure there should be sizable improvements and for most workloads. "Basically cleancache replaces I/O with memory-copy-CPU-overhead; on older single-core systems with slow memory-copy speeds, cleancache has little value, but in newer multicore machines, especially consolidated/virtualized machines, it has great value."
As part of the push into the linux-2.6.git tree today, there is CleanCache support for the EXT3, EXT4, Btrfs, and OCFS2 file-systems. Other file-systems will also likely be supported in the future. Within the Xen area is also a CleanCache shim to the Transcendent Memory.
It should come as no surprise, but Linux CleanCache benchmarks will be published on Phoronix quite soon.
13 Comments