Abstract
For the purpose of preserving large amount of increasingly digital data, many academic institutions do research on Institutional Repository (IR). IR not only prevents important research results from damaging but also increases their visibility. To maintain a long-term data preservation system might costs a lot of resources and money, so many countries put much effort on how to store digital data permanently, especially on fault tolerance and data security. In Taiwan, academic institutions preserve their digital assets separately. Doing so made preservation inefficiently, moreover it cost lots of waste. We setup a distributed data preservation system using data grid technology to gather digital files from each institution, providing consistent user interface and stable backup mechanism. This system has run for over 5 years, experiencing system upgrade and architecture modification. From analyzing system logs of one year from deferent aspects, we find some characteristics and shortcoming of the preservation system. After knowing that this system has some potential weak points, monitoring system is more and more important. So we optimize the original monitoring system, making an efficient monitoring environment, elevating stability and finally, maintaining the data preservation system in long term.