1A Fast Retrieval Method for Local or Distributed Data

Abstract—In this paper, we propose an improvement to an approach to data retrieval which is performed in only one access to a bucket hash table or file. The idea behind it, is to let the system assign one digit to the record key so that the hashed new record key is "forced " to fall in a bucket according to some practical criteria. From a user point of view this forced hash procedure could be thought of as a “user-system cooperating code assignment”, since the user is free to code an object to be retrieved but the system may append s a digit to that code. For one access retrieval purposes, the new code key-digit is used to find its address. However, should the digit is not known, the retrieval process will find the key in its surrounding, provided it exists. In this approach it is unnecessary a bucket overflow area of any kind, since this method allows a high load factor for practical use. In the event of the hash table is nearly full, a simple procedure could be ran to extend the table size either by keeping the original digit or assigning new ones. For distributed data sets this methodology shows an appealing performance in real life and simulation results