Data writing on chunks

Hi, IDS 11.10 FC3 on solaris 10. I have 4 chunks for all table data in datadbs and 1 chunk for indexes in indexdbs. out of the 4 chunks of datadbs, in 3 chunks the free pages are almost exausted and only the 4th chunk is having substantial amount of free space.. I want to know if this can affect the performance of writing the data? All the chunks are under same slice "/db/...". AFAIK If the data is distributed on different physical disks then multiple threads (multiprocessor server)can be utilized for writing the data improving the performance but will some thing like this also apply to a single chunk writing? Should adding another chunk (not for more space) be a good idea considering the performance factor? Regards, Vikas

IDS allocates extents to table using the free block with the smallest chunk number and offset within the chunk that has sufficient free space and prefers the free block closest in size to the requested extent. This all means that a newly added, completely empty, chunk will not be used by table or index extents unless either a) there is no free block big enough to hold the requested extent in the previously added chunks or b) the free space in the new chunk is the smallest free block that will accomodate the requested extent. So, adding a chunk will not affect performance significantly either way. The effect of many extents on query performance has more to do with the locality of the data within each table's extents, how many extents the table has, how spread over the chunks these extents are, and what the data access patterns of your applications is for each table. A table that is accessed such that only the most recently inserted rows are frequently accessed, and which does not experience large numbers of deletes (so that the most recent rows are on adjacent pages within a single or small number of extents) will be little affected by the physical layout of the underlying disk structures, either as to the number fo chunks, the distribution of the chunks over the disk partitions, or the number of underlying spindles making up that partition. Tables whose actively accessed data rows is spread out more will be affected more profoundly by such considerations. You have to know your data and your application access patterns. Art S. Kagel Oninit (www.oninit.com) IIUG Board of Directors (art@iiug.org) Disclaimer: Please keep in mind that my own opinions are my own opinions and do not reflect on my employer, Oninit, the IIUG, nor any other organization with which I am associated either explicitly or implicitly. Neither do those opinions reflect those of other individuals affiliated with any entity with which I am affiliated nor those of the entities themselves. On Thu, Jun 25, 2009 at 11:14 AM, VIKAS HIVARKAR <vikas.hivarkar@gmail.com>wrote: > Hi, > > IDS 11.10 FC3 on solaris 10. > > I have 4 chunks for all table data in datadbs and 1 chunk for indexes in > indexdbs. > > out of the 4 chunks of datadbs, in 3 chunks the free pages are almost > exausted > and only the 4th chunk is having substantial amount of free space.. > > I want to know if this can affect the performance of writing the data? > > All the chunks are under same slice "/db/...". > > AFAIK If the data is distributed on different physical disks then multiple > threads (multiprocessor server)can be utilized for writing the data > improving > the performance but will some thing like this also apply to a single chunk > writing? > > Should adding another chunk (not for more space) be a good idea considering > the performance factor? > > Regards, > Vikas > > > > ******************************************************************************* > Forum Note: Use "Reply" to post a response in the discussion forum. > > --001636c5a9fffb0e42046d2de508