Browse Prior Art Database

Method for fault tolerance in nonvolatile storage

IP.com Disclosure Number: IPCOM000042269D
Publication Date: 2005-Feb-03
Document File: 5 page(s) / 169K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is a method for fault tolerance in nonvolatile storage. Benefits include improved system security, improved functionality, improved performance, improved reliability, and improved yield.

This text was extracted from a Microsoft Word document.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 48% of the total text.

Method for fault tolerance in nonvolatile storage

Disclosed is a method for fault tolerance in nonvolatile storage. Benefits include improved system security, improved functionality, improved performance, improved reliability, and improved yield.

Background


              Conventionally, a single flash is found in most systems, including servers, desktop clients, and external storage boxes. System software includes startup programs, such as built-in operating system (BIOS), are preloaded into the flash. During the boot sequence, the flash contents are mapped to dynamic random access memory (DRAM) and executed. On servers, when the initialization is over, the operating system (OS) is booted. If the flash contents are corrupted, the system does not come up because the initialization sequence fails. Typically, a checksum for the entire contents of flash is included to indicate if the flash content is corrupted or not. If the contents are determined to be corrupted, the processor stops and the system halts.

      In external storage systems, the flash contains not only the initialization code but also an embedded operating system and application stack to support required protocols. If the flash is corrupted, the storage subsystem does not boot up and the entire storage underneath is inaccessible.

      Fault-tolerance for flash is required due to wear-leveling issues, such as only sustaining some number of read/writes to a given region. Additionally, flash is susceptible to inadvertent bit transitions as the line widths of semiconductor processes become successively smaller.

      Conventionally, errant flash memory causes a failure in the overall system.

      Redundant disk arrays are conventionally used to improve the reliability of hard drives that are susceptible to failure. Arrays provide sufficient hardware to enable the successful recovery  without lengthy system unavailability (see Figure 1).

      In a redundant disk array configuration, the blocks of data are interspersed among multiple drives and the parity of data blocks is maintained. The parity information is used to recover any data loss. If any one of the drives is nonfunctional, using the parity information and the data from the other drives, the contents of nonfunctioning drive can be built. The system remains operational even if one drive in the configuration is nonfunctional. Conventionally, redundant disk arrays do not include flash data volumes.

General description

              The disclosed method is fault-tolerance and fault-resilience using semiconductor nonvolatile memory, such as flash.

      The key elements of the method include:

•             Striping

•             Mirroring

•             Parity sets found in disk arrays

Advantages

              The disclosed method provides advantages, including:

•             Improved system security due to using compression logic and data encryption on the storage volume to protect the flash contents

•            ...