dbanalys in 10.1B

[maday15]

Hello! I recently upgraded from 10.0B to 10.1B (on AIX) and have had issues with the dbanalys report ever since. I think a lot of it has to do with the chain analys being added, I would only get information from some of the tables in the db and the index analys wouldn't give me any information, this past weekend it actually affected the application due to locking, I assume all this is related. I know I can use -scan which should help so I'm fine with that. The question I have now is, I noticed today that my scatter factor on a lot of my tables have gone down to under 3.0 (i had quite a few over 3 before the upgrade), is this normal after an upgrade or does 10.1B calculate the scatter factor differently? I can definitely trace the change from the upgrade, ei. Sunday before upgrade about 20 tables with scatter factor above 3.0 the day of the upgrade only 1 table has a scatter factor over 3.0.

Thanks

 

[TomBascom]

Did the "upgrade" involve a dump & load?

10.1B is certainly better than 10.0B but 10.2B06 is the current release (and has many, many significant improvements over both 10.0 and 10.1). Have you perhaps misreported the new version?

 

[maday15]

No, I didn't perform a dump and load and no I didn't misreport the version. We definitely went from 10.0B to 10.1B, we are tied to the application and can only upgrade the database to the level that matches the application. Thanks

 

[TomBascom]

I would let the vendor of that application know that I am extremely unhappy about being tied to an obsolete and unsupported release of Progress and I would escalate that to the highest levels. I would make it extremely clear that being on a release that Progress no longer supports and for which there will be no bug fixes is NOT the sort of service and support that I am paying that vendor maintenance fees for and that they should get off their butts and allow me, their customer, who makes their paycheck possible, to upgrade to a supported version of Progress. And if they make any ridiculous excuses about "certification" and other FUD regarding new Progress releases you should heap scorn upon them and suggest that that is exactly what you are paying them a rather extravagant amount of money to be taking care of. If they are not going to do so then perhaps they should refund the maintenance fees that you have been paying them. Etc.

As for the scatter factor issue... there have been some changes made in the calculation from time to time. I don't have a list of which versions changed but given that there were other obvious changes in the release that you're looking at it seems reasonable to think that that is one of those releases.

Is your data in type 1 storage areas or type 2 storage areas?

 

[maday15]

Tom, I agree. The problem is it's not entirely the vendors fault, our business unit says they don't have time to upgrade to the newer version of the application which would support newer versions of the DB. As for type of storage areas, they are Type 2.

 

[tamhas]

Point being that the existing version of the application will almost certainly run on the newest version of Progress. The vendor is just being lazy in not supporting that.

 

[Rob Fitzpatrick]

<snip> our business unit says they don't have time to upgrade to the newer version of the application which would support newer versions of the DB. </snip>

I assume you mean "newer versions of the runtime/compiler"? I don't think clients should know or care whether the back end is 10.1B or 10.2B.

What kind of clients do you have; SQL? ABL self-service? ABL remote? AppServers? Other? Are you now on the 64-bit database? What version of AIX is this?

I noticed today that my scatter factor on a lot of my tables have gone down to under 3.0

Scatter factor in Type II storage isn't particularly meaningful. I believe it is a calculation of the number of blocks your data is stored in, compared to the ideal (i.e. minimum) number that it could be stored in.

 

[tamhas]

I'm guessing he means application ... i.e., one of those ISVs who only supports older versions of Progress with older versions of their software and demands upgrading to the current version of their software to use current versions of Progress, aka too lazy to test.