M. Roth and S. Van Horn. Database compression. ACM SIGMOD Record, 22(3):31--39, September 1993.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....However, like the scheme described by Ng et al. 13] this technique is only useful for records with low cardinality fields. Westmann et al. 19] used compression at the attribute level. For numeric fields they used null suppression and encoding of the resulting length of the compressed integer [17]. For strings they used a simple variant of dictionary based compression. This is particularly e#ective if a field can only take a limited number of values. For example, a field that can only take the values male and female could be represented by a single bit which could then be used to look ....

M. Roth and S. Van Horn. Database compression. ACM SIGMOD Record, 22(3):31--39, September 1993.

....techniques. A function library, zlib, makes its functionality available to applications. We used zlib in XMill, and will refer to zlib and gzip interchangeably in the paper. Special Purpose Compressors A variety of specialpurpose compressors exists, ranging from ad hoc to highly complex ones [2, 16]. Special data types can be encoded in binary, e.g. integer or date. A dictionary encoding assigns an integer to each new word in the input, and stores the mapping from codes to strings in a dictionary. Specialized compressors exist for a variety of data types, e.g. images, sound or DNA sequences ....

....which are best compressed by specialized semantic compressors. XMill supports three kinds of semantic compressors: atomic, combined, and user defined. Atomic semantic compressors: There are eight such compressors in XMill, shown in Table 1. We explain here just a few and refer the reader to [16] or standard textbooks [17] for a general discussion. 6 The text compressor t does not compress, but rather copies the string to the container unchanged (it will be compressed later by gzip) Positive integers (compressor u) are binary encoded as follows: numbers less than 128 use one byte, those ....

[Article contains additional citation context not shown here]

M. A. Roth and S. Van Horn. Database compression. ACM SIGMOD Record, 22(3):31--39, Sept. 1993.

....an architecture for wrappers to either enforce or monitor applications on a given platform. Macoridis [24] describes wrappers that mediate between di#erent security architectures (e.g. Unix and corba) There has also been considerable work in the database community on wrapper generation. Garlic [20] and 6 We are currently working on wrapping GDB using cawom. 7 Due to space considerations, our survey of related work is representative, rather than comprehensive. 8 TSIMMIS [6] are systems that integrate heterogeneous data sources using a mediator based approach. Wrappers (called ....

M. T. Roth, M. Arya, L. M. Haas, M. J. Carey, W. Cody, R. Fagin, P. M. Schwarz, J. Thomas, and E. L. Wimmers. The Garlic Project. ACM SIGMOD Record, 1996.

....results enhances the usability, functionality and or performance of the application. 1.1. Summary of contributions Data compression has traditionally been applied to database indexing structures [WAG73, COM79, GRS98] While there has been some work on compression techniques for query evaluation [RH93, RHS95, GRS98, GS91], this activity is typically restricted to the internals of decisionsupport systems. There is also much existing research in the data compression community, mostly focused on compression algorithms for specific data types (like text and multimedia) Three issues make it non trivial to apply ....

....databases [BAS85, EOS81] In commercial database products, SYBASE IQ [SYB98] uses a compression technique similar to LZ77. DB2 [IW94] also uses Ziv Lempel compression method in the storage layer. Other researchers have investigated the effect of compression on the performance of database systems [RH93, RHS95, GRS98]. GS91] discusses the benefits of keeping data compressed as long as possible in query processing. Our research is complementary to the above work because we focus on the compression of query results leaving the DBMS. To the best of our knowledge, there is no research that directly addresses this ....

[Article contains additional citation context not shown here]

M. Roth and S. Van Horn. Database compression. ACM SIGMOD Record, 22(3): 31-39, Sept., 1993.

....techniques. A function library, zlib, makes its functionality available to applications. We used zlib in XMill, and will refer to zlib and gzip interchangeably in the paper. Special Purpose Compressors A variety of specialpurpose compressors exists, ranging from ad hoc to highly complex ones [2, 16]. Special data types can be encoded in binary, e.g. integer or date. A dictionary encoding assigns an integer to each new word in the input, and stores the mapping from codes to strings in a dictionary. Specialized compressors exist for a variety of data types, e.g. images, sound or DNA sequences ....

....codes, which are best compressed by specialized semantic compressors. XMill supports three kinds of semantic compressors: atomic, combined, and user de ned. Atomic semantic compressors: There are eight such compressors in XMill, shown in Table 1. We explain here just a few and refer the reader to [16] or standard textbooks [17] for a general discussion. 6 The text compressor t does not compress, but rather copies the string to the container unchanged (it will be compressed later by gzip) Positive integers (compressor u) are binary encoded as follows: numbers less than 128 use one byte, those ....

[Article contains additional citation context not shown here]

M. A. Roth and S. Van Horn. Database compression. ACM SIGMOD Record, 22(3):31-39, Sept. 1993.

....to applications. We used zlib in XMill, and will refer to zlib and gzip interchangeably in the paper. 9 The reversal of initials is a historical mistake, according to [4] 7 Special Purpose Compressors A variety of special purpose compressors exists, ranging from ad hoc to highly complex ones [4, 29]. Special data types can be encoded in binary, e.g. integer or dates. A dictionary encoding assigns an integer code to each new word in the input, and stores the mapping from codes to strings in a separate dictionary. Differential encoding , or delta encoding, is useful for numeric data with small ....

....semantic compressors. We distinguish three kinds of semantic compressors: atomic, combined, and user defined. Atomic semantic compressors: There are eight such compressors in XMill, shown in Table 1. We explain them next. For a more general description of semantic compressors, we refer to [29] or standard textbooks [30] The text compressor t does not compress, but rather copies the string to the container unchanged (it will be compressed later by gzip) Positive integers (compressor u) are binary encoded as follows: numbers less than 128 use one byte, those less than 16384 use two ....

[Article contains additional citation context not shown here]

M. A. Roth and S. Van Horn. Database compression. ACM SIGMOD Record, 22(3):31--39, Sept. 1993.

....results enhances the usability, functionality and or performance of the application. 1. 1 Summary of Contributions Data compression has traditionally been applied to database indexing structures [WAG73, COM79, GRS98] While there has been some work on compression techniques for query evaluation [RH93, RHS95, SNG93, GRS98, GS91], this activity is typically restricted to the internals of decision support systems. There is also much existing research in the data compression community, mostly focused on compression algorithms for specific data types (like text and multimedia) Three issues make it non trivial to apply ....

....databases [BAS85, EOS81] In commercial database products, SYBASE IQ [SYB98] uses a compression technique similar to gzip. DB2 [IW94] also uses Ziv Lempel compression method in the storage layer. Other researchers have investigated the effect of compression on the performance of database systems [RH93, RHS95, SNG93, GRS98]. GS91] discusses the benefits of keeping data compressed as long as possible in query processing. The performance of TPC D benchmark on compressed database is presented in [WKHM98] Our research is complementary to the above work because we focus on the compression of query results leaving the ....

[Article contains additional citation context not shown here]

M. Roth and S. Van Horn. Database compression. ACM SIGMOD Record, 22(3): 31-39, Sept., 1993.

....data. Section 5 discusses our TPC D results. Section 6 contains our conclusions. 2 Related Work Most related work on database compression has focussed on the development of new compression algorithms or on an evaluation of existing compression techniques for database systems (e.g. Sev83, Cor85, RH93, IW94, ALM96, NR95, GRS98] Our work differs from all this work in two important ways: First, we were interested in showing how compression could be integrated into a database system rather than inventing new compression algorithms. Second, we were interested in the performance aspects of ....

....all fields of type decimal in our implementation of the TPC D benchmark. Dictionary based and numeric compression are described below) 3.2. 1 Numeric Compression The technique we use to compress integers is based on null suppression and encoding of the resulting length of the compressed integer [RH93] This technique has also been built into ADABAS, a commercial relational database system by Software AG [AG94] The idea is to cross out leading 0 s of the representation of an integer. In most systems, integers are represented using four bytes so that Integer 3 is represented by 30 bits that ....

M. Roth and S. Van Horn. Database compression. ACM SIGMOD Record, 22(3):31-- 39, September 1993.

....techniques of knowledge discovery in databases are recently been developed. These are learning procedures which helps to find some connections among attributes, for instance, 90 percent of the consumers of product A, is a teenager with high studies ( 10, 8] On the other hand, data compression [7, 12] is also an important issue in large databases. It allows the compactification of information, reducing the number of tuples in the original database, so that the ammount of storage space can be decreased, and the eficiency in querying the database can be increased. A trivial case appears when a ....

M. A. Roth and S. J. Van Horn. Database compression. ACM SIGMOD Record, 22(3):31, September 1993.

No context found.

M. Roth and S. Van Horn. Database compression. ACM SIGMOD Record, 22(3):31--39, September 1993.

No context found.

M. Roth and S. Van Horn. Database compression. ACM SIGMOD Record, 22(3):31--39, September 1993.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC