|
|
OmniTraak offers network storage product developers fast, feature-rich, enterprise-level software components that can provide the foundation for next generation NAS storage systems. OmniTraak's simplified interfaces were designed to allow porting of this journal file system to a variety of OS platforms with a minimal engineering effort. A NetBSD version of the file system is available now. Additionally, OmniTraak's extensible structure will allow developers to quickly add high-value features or create new scalable configurations.
|
OmniTraak Components
|
|
- File System (OTFS)
- NFS server (OTNFS)
- Volume Manager (OTVM)
|
|
OmniTraak File System
|
|
Based on a 10 year history of enterprise-level service without a single report of data corruption OTFS is the heart of OmniTraak. Employing a fixed 4K-byte page size OTFS is the first file system to dynamically select either 32-bit, 48-bit, or 64-bit addressing and can support up to 64 billion terabyte volumes. Small volumes (less than 16TB) automatically default to 32-bit addressing while larger volumes will dynamically move to 48 or 64-bit addressing, optimizing efficiency, performance, and allowing open-ended growth.
|
|
Journaling and Rapid Restart
|
|
OTFS is fully journaled, meeting enterprise requirements that mandate journaling protection of both metadata and user data. Since insures data integrity in the event of power or other failure, fsck or chkdsk type routines need not be run allowing restarts in seconds regardless of volume size.
|
|
Inode Innovations
|
|
OTFS employs inode innovations designed to allow developers to add high-value features with a minimum of effort. All file systems have inodes, or something similar. Until recently file system designers studiously attempted to minimize inode size (typically 64 or 128 bytes), which often resulted in designs unable to accommodate unforeseen future requirements. Additionally, the preallocated inode tables were invariably larger than required resulting in wasted disk space. Acknowledging that small, handcrafted inode designs can prove brittle and are ill prepared to accommodate emerging requirements Traakan adopted a more flexible approach that allows the addition of new features without disruption of an installed base. OTFS also avoids awkward, performance-sapping workarounds like side files.
Designed for maximum flexibility and optimized for performance OTFS inodes are stored in dynamically allocated pages like other OTFS data structures. These inodes can support variable numbers of "properties" with varied length and shape, dynamic 32/48/64-bit addressing, and can easily handle future ACLs extensions. Additionally, inode pages can often accommodate a file's complete structure including its attributes, ACLs, and content. Since typically 50% of all files are small enough to be included in a single page, OTFS inodes can significantly improve disk utilization and enhance performance by reducing subsequent disk seeks.
|
Enterprise Capabilities
|
Snapshots and BRAM
|
|
Snapshots are virtual, instantaneous, read-only copies of an entire file volume. While the file volume may continue to change, the snapshot view appears unchanged. This stabilizes backups and allows users to do unassisted recovery of mistakenly modified or deleted files. OTFS supports up to 240 snapshots with negligible overhead.
Backup and Replication Access Method (BRAM): identifies and stabilizes pages that have changed since a snapshot. An OmniTraak first, applications can access BRAM using ordinary file interfaces.
|
|
Performance Enhancements
|
|
Automatic Tuning: As the mix and load of file operations change, OTFS dynamically tunes parameters to achieve the best overall response time.
Fast Hashed Directories: A directory of 100,000 entries delivers the same performance as a traditional directory of 1,000 entries.
Oversized Main Memory: Memory sizes in excess of 4GB are supported. They are used for large data buffers, providing improved performance.
|
|
OmniTraak NFS
|
|
The NFS module (OTNFS) intercepts NFS requests early in the network stack and routes them directly to OTFS. This removes unnecessary processing and queuing overhead that is associated with the general-purpose socket layer and the VFS abstraction layer. While entirely compatible with VFS or similar methods, the high-speed bypass directly to OTFS saves considerable processor power and boosts performance. This module, in conjunction with OTFS, has successfully been interoperability tested with major storage vendors over the past ten years.
|
|
OmniTraak Volume Manager
|
|
OTVM pools (or "virtualizes") disk storage then portions it out to logical volumes. The storage can be directly attached (ATA, SCSI), Fibre-Channel attached (SAN), or attached through standard networking (iSCSI). Users are presented with a single storage pool and can quickly create new volumes or grow existing ones. Whether employing a few simple disks or large amounts of RAID storage, OTVM greatly simplifies management and deployment of storage resources.
|
|
|
