| To tackle the defects such as the lack of IP address and poor security exposed by TCP/IP network,a novel network architecture named Named Data Networking is proposed.Named Data Networking supports the efficient sharing and distribution of data by providing in-network caching capability via built-in Content Store.Retrieving and storing massive data brings great challenges to the design and deployment of Content Store.Our work focus on studying the storage structures and name lookup algorithms of Content Store to design the enhanced and deployable Content Store.Some innovative works are carried out below.Firstly,a novel Content Store named LT-DSL is proposed based on Double Skip List(DSL)to meet the requirement about supporting the cache replacements.LT-DSL consists of two parts: Learning Tree and DSL.LT-DSL can realize the fast name lookup and reduce the memory consumption by redesigning the index and the name lookup algorithms.The experiment results demonstrate that LT-DSL can support the cache replacements,promote the lookup speed,and further minimize the memory consumption under the premise of meeting the requirements of current network.Secondly,an improved Content Store named LLT is proposed based on Logical Table to meet the requirement about supporting the All Sub-Name Matching(ASNM)and the Exact Name Matching(ENM)quickly.LLT consists of two parts: Learning Tree and Logical Table.LLT can support the ASNM and the ENM,realize the fast name lookup and reduce the memory consumption by moving the lookup forward to the insertion process.The experiment results show the effectiveness and efficiency of LLT: under the premise of effectively supporting more complex name lookup algorithms and meeting the requirements of current network,LLT can reduce memory consumption and achieve 177.37 MPPS single-thread throughput. |
