Data Reliability

DATA PROTECTION AND RECOVERY

With explosive growth of networked information services and e-commerce, data protection and recovery have become the top priority of business organizations and government institutions. Since data is the most valuable asset of an organization, any loss or unavailability of data can cause millions of dollars of damage . Unfortunately, failures do occur such as hardware failures, human errors, software defects, virus attacks, power failures, site failures, and so forth. In order to protect data from possible failures and to recover data in case of a failure, data protection technology is necessary.

Traditionally, data protection has been done using periodical backups. At the end of a business day or the end of a week, data are backed up to tapes. Depending on the importance of data, the frequency of backups varies. The higher the backup frequency, the larger the backup storage is required. In order to reduce the backup volume size, technologies such as copy-on-write (COW) snapshots and incremental backups have been commonly used. Instead of making full backups every time, COW snapshots and incremental backups that only store the changed data are done more frequently in between full backups. For example, one can do daily incremental backups and weekly full backups that are stored at both the production site and the backup site. In this way, great storage savings are possible while keeping data protected.

Despite the importance of data protection and recovery, recent study has shown that 67% of backup data cannot be recovered in the real world. Even if data can be recovered, it takes hours and even days to do so. While this fact is well known, there has been no research study on why this is the case. Therefore, it remains unclear and an open question why such high percentage of data recovery failed.

We carry out theoretical study as well as experiments for data protection and recovery. Recent findings include:

Upset about damaging your phone and losing your contacts, notes from meetings and Angry Birds high score? Fear not, engineering Professor Qing (Ken) Yang recently patented a process to recover data from damaged flash memory, the storage component of most portable devices. Relying on an innovative algorithm and hardware-software interface, Yang’s techniques will bring enhanced data recovery options to consumers and criminal investigators searching for evidence.

What does “TRAP” stand for? It is an abbreviation of “Timely Recovery to Any Point-in-time”