Research On Application Of Network Processor In Data Channel For Holographic Storage System

There are two main types of storage media. One is magnetic, the other is optical.Magnetic is the earliest media found for external storage. It included from the early magnetictapes to later magnetic disks, which are widely used in main frames and small computersystems. On the other hand, the compact disk came out in 70s and thus began the history ofoptical storage. Today CDR, DVDR, DVDRW are drawing more and more spotlight on thestage. However, the current storage medium in use, either magnetic or optical, istwo-dimensioned. The data was stored on the surface of the media, which limits the capacityand row speed. Scientists have been working on volume holographic storage, thecharacteristic of which is to store and retrieve data on the interference pattern. Multiplepatterns can be accessed at a single storage volume section. The result is much highercapacity and much higher data speed. Thus, it becomes highly necessary to research on thehigh speed data interface and data channel to accommodate the new storage media.After comparison, the IP Storage design goal is set because the IP based storage system hasvery flexible design capability and can be of broadest field of application. To achieve hightransfer rate, the latest network processor technology is used to build the transfer engine. TheIXP 1200 network processor is chosen to be the data engine for its flexibility inprogramming. There are four task modules for the target system, most important of which to the outerdata channel is the rebuild of the basic function of IXP1200 in the system. First a mostsimilar reference design designated for packet exchange is chosen. And then the softwarearchitecture and data flow is analyzed. The software frame is then broken in order to strip theIP routing module unnecessary to the target system. New data structures are designed toaccommodate the PCI module to the architecture. Thus, the software architecture and basicfunction of the IXP1200 network processor is turned from packet exchange to data requestprocessing and transferring. The key technology is first to delicately break the internaltransferring step to make the PCI processing module seamlessly merge into the receiving andtransmitting modules. The most crucial part is the design and integration of the internal datastructures of the PCI module. Based upon the above work the more detailed designingblueprint and implementation method is given, too.