× The internal search function is temporarily non-functional. The current search engine is no longer viable and we are researching alternatives.
As a stop gap measure, we are using Google's custom search engine service.
If you know of an easy to use, open source, search engine ... please contact support@midrange.com.



The performance improvements occur in rather narrowly defined regions. E.g., the assumption that there are many more logical reads than writes might not be the case for you. You can verify this using DSPFD, which lists these IO statistics. I recommend verifying the nature of your IO activity before jumping to DDL-defined tables. Centerfield Technology, my just previous employer, has a database assessment that takes the historical IO counts into consideration when estimating the impact of adding indexes. This is, I think, an important analytical step to take before doing as suggested in the article.

Also, IMO, Figure 5 is somewhat misleading - the point is still made, but the scales on the 2 sides are different. I believe the green line should go from top to top of the 2 stacked bars.

Also, I noticed, in the paragraph between the references to figures 4 & 5, that he apparently did not create an SQL index, it says he created a logical file over the SQL table. This skews things even further.

Figure 7 suffers from the same distortion as Figure 5 - the scales are not the same.

We also do not know, do we, what the expert cache settings were? One needs to use WRKDSKSTS and mayb WRKSYSACT to find out what the physical transfer sizes and rates have been. Different patterns of data access (sequential vs. random, say) can result in vastly different performance. Of course, if the pool is set to *FIXED (not recommended), then everything is at least on a more level playing surface - probably.

The article brings up the matter of REUSEDLT, which is *NO by default with physical files. Setting this to *YES will probably have some beneft, since there will not be as many *DELETED records to go through. But it should not be done if there is any dependency on the arrival sequence of the data. Granted, that is an older technique and will not be likely to be used with SQL-type objects. But be careful.

And, of course, the performance gain with REUSEDLT really has nothing to do with SQL-generated tables, other than the changed default. Shops that did not take advantage of this could have long ago, if otherwise appropriate.

Q.E.D. - I do not think we can rely on these tests only, to take action - sorry to say that, Folks looking at this kind of change need to do more analysis of their particular application and DB I/O mix. I do not know the author at all, and it is easy to criticize this type of thing from afar, esp. when I don't have the time to try much out myself. But I am very ready to hear from said author, or anyone else, and to discuss these things.

Vern

At 05:25 PM 7/5/2005, you wrote:



> but it seemed to me that developers could use REFACCPATH directives in DDS, or record blocking directives in RPG, or other available techniques to swing the performance results in favor of files created via DDS.

Nathan,

the "selling point" of this article is that it offer a incredible performance boost (43%) without any development (sometimes is just a time&cost, sometimes it's even impossible to modify 3rd party applications).

Dmitri Efimov


Aviso Banco Wiese Sudameris

Este mensaje es privado y confidencial y solamente para la
persona a la que va dirigido. Esta sujeto a los terminos que
aparecen en  http://www.bws.com.pe/aviso.shtml los cuales
forman parte de su contenido.

--
This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing list
To post a message email: MIDRANGE-L@xxxxxxxxxxxx
To subscribe, unsubscribe, or change list options,
visit: http://lists.midrange.com/mailman/listinfo/midrange-l
or email: MIDRANGE-L-request@xxxxxxxxxxxx
Before posting, please take a moment to review the archives
at http://archive.midrange.com/midrange-l.


As an Amazon Associate we earn from qualifying purchases.

This thread ...

Replies:

Follow On AppleNews
Return to Archive home page | Return to MIDRANGE.COM home page

This mailing list archive is Copyright 1997-2024 by midrange.com and David Gibbs as a compilation work. Use of the archive is restricted to research of a business or technical nature. Any other uses are prohibited. Full details are available on our policy page. If you have questions about this, please contact [javascript protected email address].

Operating expenses for this site are earned using the Amazon Associate program and Google Adsense.