Skip to main content

Hitachi

Corporate InformationResearch & Development

December 6, 2013

Report from Presenter


Photo 1 Best paper award

The 2nd Asian Conference on Information Systems (ACIS 2013) was held in Phuket, Thailand, from Oct. 31 to Nov. 2, 2013. Many researchers, professionals and practitioners discuss various current issues on information systems in ACIS, and the number of participants is growing every year. In this year, there were 76 presentations.

We, Yokohama Research Laboratory, Hitachi, Ltd., made two oral presentations in ACIS 2013, and one of them was awarded as one of the best papers. This report shows the awarded one, entitled "Risk-aware Data Replication to Massively Multi-sites against Widespread Disasters" (Photo 1).

There are important information services to sustain even immediately after a widespread disaster, such as an earthquake and tsunami. For example, it is necessary to service census registers for local governments to identify people and to check whether people are safe or not. For another example, it is necessary to service medical records and prescriptions for medical institutions to give appropriate medical care.

To sustain such important services, we often utilize a Disaster-Recovery system (DR system), which enables the service to sustain immediately after a disaster by replicating necessary data to a remote backup site in advance. However, the Great East Japan Earthquake of 2011 taught us that a DR system included a problem in the case of widespread disasters. The problem is that people in a disaster area cannot access the service provided by a backup site in a DR system. This problem is caused because a widespread disaster damages a wide-area network which connects people in a disaster area to the remote backup site (Fig. 1).


Fig. 1 Problem in data protection system
Enlarge


Fig. 2 Overview of Risk-aware Data Replication
Enlarge

We propose Risk-aware Data Replication (RDR) to solve this problem. RDR replicates data not to a conventional remote backup site but to multiple "safe" backup sites in the neighborhood. To find out "safe" backup sites against a widespread disaster, we modeled the data loss risk and utilized Mathematical Programming technique. By doing these, RDR can seek the safe backup sites systematically even if the number of the sites is massive (Fig. 2). Simulation results show that RDR is possible to protect more than 20% amount of data than the method which seeks backup sites just in random manner.

This research is supported as "Research and Development on Highly-functional and Highly-available Information Storage Technology", sponsored by the Ministry of Education, Culture, Sports, Science and Technology in Japan, and we are collaborating and researching with Tohoku University and Hitachi Solutions East Japan, Ltd. We will continue researching this issue, implement RDR to a storage system which is underlying a simulated medical information service, and evaluate the usefulness in 2016.

(By MATSUMOTO Shinya)

  • Page top