BlueArc's patented hardware-based file system provides the performance and scalability to meet the demands of today's NAS requirements, without the legacy of slower software-based file servers.
Unlike traditional file servers that use common operating systems and file systems on standard PC-based hardware, BlueArc started from the ground up to develop a unique hardware architecture that does not simply accelerate the file system in hardware, but actually puts the operating system and file system into hardware, literally into silicon. Titan uses programmable logic chips and patented engineering methodology to design an object-based operating system and file system into many of these programmable chips, each with specific state engines and memory to perform specific tasks. This allows massive parallelism in handling all types of requests as many requests such as network, file system, or storage requests can all be handled simultaneously in one clock cycle. This is similar to network switches and routers that use hardware and a slim operating system to provide high performance packet movement.
BlueArc's unique file system and operating system is purpose built and thus does not have the bloat of software-based systems. Also as the hardware is designed with future growth in mind, there is no slow-down as features are added. A simple way to see the difference between the software and hardware architectures is in how directories are handled. In a software-based filer, as directories fill up and response gets sluggish. Although limitations are often in the 10's of thousands, reality is that administrators know to keep these numbers much lower. They often create complex folder structures to limit the number of files in a folder. BlueArc's hardware-based file system can handle up to 4 million files per directory, and does not suffer the slow-down of the software filers as the directory is kept as a binary lookup and all queries are handled in hardware quickly and efficiently. This type of flexibility is critical in areas such as Internet Services or Life Sciences where they may have billions of small files to store.
The file system itself is an objected-based file system, where an object is one or more raw blocks in a tree structure. Each element or an object is called an Onode. The primary element is called the Root Onode, and has a unique 64-bit Object Identifier (OID) and metadata relevant to it. These Root Onodes can point directly to Data Onodes or to pointer Onodes depending on the amount of content to be stored. Via this extensibility the file system can handle a single object as large as the entire file system or billions of smaller objects in an equally efficient manner. Objects are manipulated by the logic residing within the programmable logic on the file system module of Titan. Functions like Virtual Volumes and Snapshots are built into the hardware architecture, to ensure the highest levels of performance even during data protection activities.
For client servers or users, they have no concept of objects. Although unique on the inside, Titan is designed with standards in mind, and all data access is accomplished using industry standard protocols such as CIFS, NFS, iSCSI, and NDMP. This eliminates the need to have specialized drivers or hardware on the hundreds or thousands of client servers, which can be costly and difficult to manage.
For a deeper dive on the Titan system and the object-based file system download the whitepaper BlueArc's Titan Architecture.
