TPT-RAID: a High Performance Box-Fault Tolerant Storage System

TPT-RAID is a multi-box RAID wherein each ECC group comprises at most one block Jrom any given storage box, and can thus tolerate a boxJailure. It extends the idea ojan out-oj-band SAN controller into the RAID: data is sent di­ rectly between hosts and targets and among targets, and the RAID controller supervises ECC calculation by the targets. By preventing a communication bottleneck in the controller, excellent scalability is achieved while retaining the simplic­ ity oj centralized control. TPT-RAlD, whose controller can be a software module within an out-oj-band SAN controller, moreover conJorms to a conventional switched network ar­ chitecture, whereas an in-band RAID controller would ei­ ther constitute a communication bottleneck or would have to also be a full-fledged roUler. The design is validated in an InfiniBand-based prototype using iSCS[ and iSER, and required changes to relevant protocols are introduced.

• Type: Conference
• Year: 2007
• Where: MSS
• Authors: Yitzhak Birk, Erez Zilber
• Keywords: Controller Supervises Ecc, Internet Technology, MSS 2007, Out-oj-band San Controller, RAID Controller