Weaving Relations for Cache Performance

By A. Ailamaki, David J. DeWitt, Mark D. Hill, Marios Skounakis. Very Large Data Bases Conference, 2001

440 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/0109b8d4f75feed4ffbd4b5d555bac1e2d27815d

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Bigtable: a distributed storage system for structured data (2006, background): “Locality groups do not support CPUcache-level optimizations, such as those described by Ailamaki [2].”
Impala: A Modern, Open-Source SQL Engine for Hadoop (2015, background): “Simply described, Parquet is a customizable PAX-like [1] format optimized for large data blocks (tens, hundreds, thousands of megabytes) with built-in support for nested data.”
Database Cracking (2007, background): “Some notable departures from the usual paths in database architecture design include [2, 3, 9, 14].”
RCFile: A fast and space-efficient data placement structure in MapReduce-based warehouse systems (2011, methodology): “RCFile is designed and implemented on top of the Hadoop Distributed File System (HDFS).”
SW-Store: a vertically partitioned DBMS for Semantic Web data management (2009, methodology): “Thus, SW-Store adopts a hybrid storage representation along the lines of PAX [8], where a single block contains data from both columns; however, within a block, data from each column are kept separate.”
Integrating vertical and horizontal partitioning into automated physical database design (2004, background): “We note that recent studies [3,16] have looked into new DBMS architectures where improved cache performance is possible by storing data in a column-wise manner.”
Fast computation of database operations using graphics processors (2004, background): “The pixel processing engines have direct access to the texture memory and can perform vector operations with (cid:13)oating point arithmetic.”
The Snowflake Elastic Data Warehouse (2016, methodology): “Within each ﬁle, the values of each attribute or column are grouped together and heavily compressed, a well-known scheme called PAX or hybrid columnar in the literature [2].”
In-Memory Big Data Management and Processing: A Survey (2015, background): “This is caused by a huge amount of data cache misses [169], which account for 50-70 percent for OLTP workloads [91], [170] to 90 percent for DSS workloads [91], [171], of the total memory-related stall.”
Parallel data processing with MapReduce: a survey (2012, unknown): “in column-wise, similar to PAX storage scheme [3].”
Integrating compression and execution in column-oriented database systems (2006, background): “This is because row-stores fundamentally are incapable of mixing attributes from more than one tuple in a single entry if other attributes of the tuples are not also included in the same entry (by definition of “row-store” – this statement does not hold for PAX-like [4] techniques that columnize blo”
MonetDB/X100: Hyper-Pipelining Query Execution (2005, methodology): “Recent database research has shown that DBMS performance is strongly impaired by memory access cost (\cache misses") [3], and can signi(cid:12)cantly improve if cache-conscious data structures are used, such as cache-aligned B-trees [16, 7] or column-wise data layouts such as PAX [2] and DSM [8] (as”
Implementing database operations using SIMD instructions (2002, methodology): “For example, there has been much recent work on CPU cache behavior in database systems [3, 6, 11, 18, 21, 22].”
Super-Scalar RAM-CPU Cache Compression (2006, methodology): “While ColumnBM by default uses the DSM storage model [6], we also present the results for PAX storage [1].”
GPUTeraSort: high performance graphics co-processor sorting for large database management (2006, background): “These techniques can also improve the performance of sorting algorithms [5, 12, 26, 28, 36, 37, 39, 45].”