OCFS2 vs RAW vs EXT 2/3
Introduction to OCGF2
OCFS2 is the next generation of the Oracle Cluster File System for Linux. It is an extent based, POSIX compliant file system. Unlike the previous release (OCFS), OCFS2 is a general-purpose file system that can be used for shared Oracle home installations making management of Oracle Real Application Cluster (RAC) installations even easier. Among the new features and benefits are:
- Node and architecture local files using Context Dependent Symbolic Links (CDSL) (symbolic link to different file on different hostname/architecture)
- Network based pluggable DLM (distributed lock management)
- Improved journaling / node recovery using the Linux Kernel "JBD" subsystem
- Improved performance of meta-data operations (space allocation, locking, etc).
- Improved data caching / locking (for files such as oracle binaries, libraries, etc)
OCFS2 comes bundled with its own cluster stack, O2CB. The stack includes:
- NM: Node Manager that keep track of all the nodes in the cluster.conf
- HB: Heart beat service that issues up/down notifications when nodes join or leave the cluster
- TCP: Handles communication between the nodes
- DLM: Distributed lock manager that keeps track of all locks, its owners and status
- CONFIGFS: User space driven configuration file system mounted at /config
- DLMFS: User space interface to the kernel space DLM
O2CB instance runs on each node, and provides the read/write and lock control for nodes access, throught network communication.
ORACLE SUPPORT AND CERTIFICATION
OCFS2 1.2.3 or later is certified with Oracle 10gR2 RAC on the x86 and x86-64 platforms on Enterprise Linux 4. Users can download the packages from the Unbreakable Linux Network site.
OCFS2 1.2.1 or later is certified with Oracle 10gR2 RAC on the x86, x86-64, IA64, PPC64 and s390x platforms on Red Hat’s RHEL4 Updates 2 and higher. Users can download the packages from here.
OCFS2 1.2.1 or later is certified with Oracle 10gR2 RAC on the x86, x86-64, IA64, PPC64 and s390x platforms on Novell’s SLES9 SP3. This release is included with all kernels starting 2.6.5-7.257 and can be downloaded from the SuSE Patch Support Database (PSDB) or by using the YaST Online Update (YOU).
OCFS2 1.2.1 or later is also certified with Oracle 9iR2 and 10g RAC on the x86, x86-64 and IA64 platforms on the above mentioned RHEL4 and SLES9 releases.
Comparison with RAW and EXT 2/3
OCGF2 is a cluster file system, but Ext 2/3 is not, some optimization towards Ext 2/3 will not be applicable to OCGF2. For example, the cache mechanism, which leverages the reading performance of ext, can’t be apply to OCGF2, as what OCGF2 want to build is an simple view of files that can grow and shrink with a clear view across all nodes.
The way OCFS does I/O is the SAME way as IO is done to a raw device inside the kernel. There is no caching at all when Oracle does an I/O to OCFS, there is no I/O through the Linux buffer cache, there is no OS caching, nothing. Every read and every write accesses the device.
For writes the nice thing here is that the kernel side of the process will actually be faster, it’s a direct path to disk, no waiting for someone else to go around flushing buffers and coming back.
Therefore: RAW and OCFS are very comparable in performance.