All added files comply with the license rules of the Linux kernel.
File system journaling critically limits the scalability of a file system because all simultaneous write operations coming from multiple cores must be serialized to be written to the journal area. Although a few scalable journaling approaches have been proposed, they required the radical redesign of file systems, or tackled only a part of the scalability bottlenecks. Per-core journaling, in which a core has its own journal stack, can clearly provide scalability. However, it requires a journal coherence mechanism because two or more cores can write to a shared file system block, so write order on the shared block must be preserved across multiple journals. In this paper, we propose a novel scalable per-core journal design. The proposed design allows a core to commit independently to other cores. The journal transactions involved in shared blocks are linked together through order-preserving transaction chaining to form a transaction order graph. The ordering constraints later will be imposed during the checkpoint process. Because the proposed design is self-contained in the journal layer and does not rely on the file system, its implementation, Z-Journal, can easily replace JBD2, the generic journal layer. Our evaluation with FxMark, SysBench and Filebench running on the ext4 file system in an 80-core server showed that it outperformed the current JBD2 by up to approx. 4000 %.
- Download and build e2fsprocs for zjournal.
git clone https://github.com/J-S-Kim/e2fsprog-zj.git
cd e2fsprog-zj
./configure
make
- Download and install the kernel of the current repository, and reboot to the kernel you installed.
- When installing, configure should be set as below.
- After rebooting, verify that ext4mj and zj modules are installed successfully.
File systems --->
[M] The Extended 4 (ext4mj) filesystem
- Format the storage device and mount it with the command shown below.
- At this point, the number of journals is the same as the current number of online cores.
sudo ./e2fsprog-zj/misc/mke2fs -t ext4 -J multi_journal -F -G 1 /dev/<block device>
sudo ./e2fsprog-zj/misc/tune2fs -o journal_data /dev/<block device>
sudo mount -t ext4mj /dev/<block device> <mount point>
- Z-Journal mainly modified transaction.c, commit.c, and checkpoint.c of the fs/zj directory, and super.c, fsync.c, mballoc.c, ext4mj_zj.c, ext4mj_zj.h of the fs/ext4mj directory.
- The contents of Journal Coherence Commit are reflected in commit.c and transaction.c.
- The contents of Journal Coherence Checkpoint are reflected in checkpoint.c
- Modification of super.c is necessary to recognize multiple journals, and ext4mj_zj files are necessary for good comfort with existing journaling APIs.
- fsync proceeds in a different way than before due to the per-core journaling. Therefore, the fsync.c file was modified.
- The mballoc file includes modification of core-aware block group allocation (BGA).
- Fileserver
- Varmail
