Pawel,
>jBASE will invoke distribution routine for each @ID. This will introduce > ***significant overhead*** (it will be much bigger if your distribution > subroutine is complex). I am not sure how is your routine written but I > hope it is 1 line maximum (and that you did not include I_COMMON :)). ok it sounds sensible.the routine is like 1 line (a programmer has done it not me,i dont know a lot of programming :)) and i think it just distributes the records based on the number of a specific digit or two of the @ID. > You could also try to force jBASE to use indexes to avoid full select > scans. We were trying recently on jBASE 4.1.5.31, but were not able to > get benefits for pure SELECT statements (KEY-SELECT worked perfectly, > but SELECT without criteria not). > I still do not know why jBASE team did not create multiple threads to > query partfiles separately avoiding distribution subroutine calls. > Perhaps some minor/major jBASE revision will bring that performance > related option? > I also guess from your post that you never archived your accounting > data. Did you ever use T24 archiving? :) > i dont think we use indexes and yes i have used Globus archiving but not for CATEG.ENTRY because for some, unknown to me, reason all the records are needed on the production environment so we distributed it. > > size of each part is roughly 610MB & modulo is 105019 and > > the recommended is 110491. > > So it seems that it is not related to sizing. > > > 2)im not sure what you mean by that so i paste the output of jstat -v > > of 1 part file: > > I was asking about "Type" and eventually hash method (I do not recognize > hash methods :)). Jim suggested likely best option (J4, hash=2). Check > this newsgroup archives to see some comparisions or try different > options by yourself :) > > > 3)i dont think we have triggers but ill have to ask a programmer about > > that. > > LIST-TRIGGER <file> will tell you wheter you use triggers or not. You > likely do not have them. They may/will introduce some overhead to data > access layer. > > > 1+2) we are using 4x 146GB 15K SAS hds 4Gbit FC on RAID 10.so they are > > basically 2 physical disks mirrored on another pair. > > these appear as 1 PV (physical volume) through AIX. > > I can not compare, neither specify our setup. We are testing new > hardware now. > > > 3) "how many ranks?"i m not familiar with the ranks term. > > I think this term is related to DS8000 only, so sorry for confusion > caused (http://www.redbooks.ibm.com/abstracts/tips0535.html?Open). I am > that big storage specialist :D yeah probably.cant find anything about ranks. > > 4)i have been looking for the stripe size for some time but i cannot > > find anything like that on the storage. you are probably asking about > > the raid stripe size of the array. (segment size is 128KB which is > > also another relevant parameter but adjusting it to 256KB some time > > ago didnt really gain a significant advantage.) > > You would need to chat to our storage manager to discuss that. I do not > know wheter there were any performance gains. > > > 5) no the SAN is used by many servers.there are a few other RAID > > arrays used by windows servers.still even when i test the system when > > im the only user on the server and SAN utilisation is fairly idle i > > get roughly the same numbers. > > And that may be the point also. External, non-T24 traffic may degrade > your performance significantly. I assume however that you did your > testing carefully (stable results). yeah the tests are quite fair as i did them when i was the only one logged on the server and mostly when the SAN was fairly idle.The SAN is used for other servers as well but i tested when there was not really any traffic for the others. > > 6) yeah IBM says that there is no problem at all with the > > storage.nevertheless they do care about the AIX performance which is > > ok, so if jbase acts weird they wont really care about it a lot. > > It does not do many weird things I think. > > > would worry a bit less :). i just want have a view about hows other > > people running in general cause i havent got really anybody else to > > compare and talk about it. > > I think that people here (Poland) could discuss onsite some things with > you. We had already other banks visiting us. There is always field of > comparison. I can propose nothing officially (it is not me), but... :) > > Kind regards > Pawel well IBM still says its ok since it works fine through AIX commands, works well for windows, and we even tried several GB copied from expansion to expansion withing the SAN through the FCs and got speeds about 505MB/sec which is impressive since its theoritical maximum is 4Gbits (512MB) Im in Greece by the way. (I have been testing some new hardware cause in about the 15days the transition will occur.) Thanks. -- Please read the posting guidelines at: http://groups.google.com/group/jBASE/web/Posting%20Guidelines IMPORTANT: Type T24: at the start of the subject line for questions specific to Globus/T24 To post, send email to [email protected] To unsubscribe, send email to [email protected] For more options, visit this group at http://groups.google.com/group/jBASE?hl=en
