Re: Online V7 and Parallel Queries
In article <329b4022.18982794@news.dircon.co.uk>, Jeremy Russell <jeremyr@dircon.co.uk> writes >jeremyr@dircon.co.uk (Jeremy Russell) wrote: > >>Is it true that PDQ, parallel queries in OnLine V7.x will use both >>primary chunks and mirror chunks to retrieve data concurrently when >>performing a query expected to return large volumes of data??? >> > >Thanks to all who responded to this, and so promptly too! In summary, >the answers I got were Yes, No and Possibly. If anyone wants to see >all the replies in detail - mail me. > > >-- >Jeremy Russell - jeremy@skilladvance.co.uk (work) > jeremyr@dircon.co.uk (home) >SkillAdvance, independent UNIX, RDBMS and Windows > training - http://www.skilladvance.com On the Online 7.x Admin COurse I went on last year the tutor was not sure as well. Seems that even some people at Informix are not sure... According to the INFORMIX-OnLine Dynamic Server Administrator's Guide, Volumn 1 Version 7.1 Page 27-9:- Disk Writes to Mirrored Chunks During Online processing, OnLine performs mirroring by executing two writes for each modification: one to the primary chunk and one to the mirror chunk. Virtual processors of the AIO class perform the actual disk I/O. For more information, refer to "Asynchronous I/O" on page 10-24. The requesting thread submits two write requests (one for the primary chunk and one for the mirror chunk) asynchronously. That us, if two AIO virtual processors are idle, they can perform the two disk writes in parallel. In the meantime, the requesting thread can perform any additional processing that does not depend on the results of the mirror I/O. Disk Read from Mirrored Chunks OnLIne uses mirroring to improve read performance because two versions of the data reside on separate disks. A data page is read from either the primary chunk or the mirror chunk, depending upon which half of the chunk includes the address of the data page. This feature is called a split read. Split reads improve performance by reducing the disk-seek time. Disk-seek time is reduce because the maximum distance over which the disk head must travel is reduced by half. Figure 27-2 illustrates a split-read. (There is then a diagram which shows a chunk as a circular band on a disk viewed from above with a circle through the middle of the chunk spliting it in two. Arrows show that the inside half is read from the primary chunk and the outside half from the mirror chunk). i.e. /-------------\\ / <----\\-----Data read from / /-------\\ \\ mirror chunk / / \\ \\ / / /---\\ \\ \\ | | | | | | Data read from ---|----|-> \\---/ | | primary chunk \\ \\ / / \\ \\---------/ / \\ / \\---------------/ -- David Williams