Daisy Seminar

User Tools

Site Tools

Trace: • 20111021
20110916

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revisionPrevious revision
Next revision		Previous revision
20110916 [2011/09/15 10:11] – simas20110916 [2011/09/16 12:25] (current) – root
Line 1: Line 1:
 20110916 DBlunch 20110916 DBlunch
  
-I will present our IDEAS 2011 paper:+News: \\ 
 +FYI Affiliation analysis of database publications 
 +http://www.sigmod.org/publications/sigmod-record/1103/pdfs/06.forum.aumuller.pdf 
 + 
 +presenters for next time:\\ 
 +Torben Bach Pedersen\\ 
 +Kasper Søe Luckow \\ 
 + 
 +Attendance: \\ 
 +  * Torben Bach Pedersen 
 +  * Kristian Torp 
 +  * Simonas Saltenis 
 +  * Ove Andersen   
 +  * Andreas Weisberg 
 +  * Saulius Samulevicius 
 +  * Laurynas Siksnys 
 +  * Benjamin Krogh 
 +  * Darius Sidlauskas 
 +\\ 
 +Presentation:\\ 
 +I(Simonas) will present our IDEAS 2011 paper:
  
 FB-Tree: A B+-Tree for Flash-Based SSDs FB-Tree: A B+-Tree for Flash-Based SSDs
Line 7: Line 27:
 Martin Vang Jørgensen, René Bech Rasmussen, Simonas Šaltenis, Carsten Schjønning Martin Vang Jørgensen, René Bech Rasmussen, Simonas Šaltenis, Carsten Schjønning
  
-Abstract:+Abstract:\\
 Due to their many advantages, flash-based SSDs (Solid-State Drives) have become a mainstream alternative to magnetic disks for database servers. Nevertheless, database systems, designed and optimized for magnetic disks, still do not fully exploit all the benefits of the new technology. We propose the FB-tree: a combination of an adapted B+-tree, a storage manager, and a buffer manager, all optimized for modern SSDs. Together the techniques enable writing to SSDs in relatively large blocks, thus achieving greater overall throughput. This is achieved by the out-of-place writing, whereby every time a modified index node is written, it is written to a new address, clustered with some other nodes that are written together. While this constantly frees index nodes, the FB-tree does not introduce any garbage-collection overhead, instead relying on naturally occurring free-space segments of sufficient size. As a consequence, the FB-tree outperforms a regular B+-tree in all scenarios tested. For instance, the throughput of a random workload of 75% updates increases by a factor of three using only two times the space of the B+-tree. Due to their many advantages, flash-based SSDs (Solid-State Drives) have become a mainstream alternative to magnetic disks for database servers. Nevertheless, database systems, designed and optimized for magnetic disks, still do not fully exploit all the benefits of the new technology. We propose the FB-tree: a combination of an adapted B+-tree, a storage manager, and a buffer manager, all optimized for modern SSDs. Together the techniques enable writing to SSDs in relatively large blocks, thus achieving greater overall throughput. This is achieved by the out-of-place writing, whereby every time a modified index node is written, it is written to a new address, clustered with some other nodes that are written together. While this constantly frees index nodes, the FB-tree does not introduce any garbage-collection overhead, instead relying on naturally occurring free-space segments of sufficient size. As a consequence, the FB-tree outperforms a regular B+-tree in all scenarios tested. For instance, the throughput of a random workload of 75% updates increases by a factor of three using only two times the space of the B+-tree.
20110916.1316081506.txt.gz · Last modified: by simas