User talk:Michel Cadot
Best regards. Frank Naude 13:26, 6 December 2008 (UTC)
Due to the number and size of the images I splitted the page in two parts.
The first part gives statistics from the beginning of the forum (but for the latest section, "Forum activity along the day", I added at the bottom in October).
The second part gives the evolution of statistics per year and statistics for the last 12, 6 and 3 months.
Links to the second page are included in the first one.
I think there are some uploaded images that are no more used but I can't find them. All used images are suffixed with ".JPG" in upper case. All images you can find with ".jpg" (in lower case) are obsolete. I discovered last time that your site distinguished between the two.
- Great, thanks Michel! Best regards. Frank Naude 07:46, 10 December 2008 (UTC)
I don't agree with the sentence you added in this article:
Note that whatever is the configuration you give, Resource Manager comes into play only if your server CPU is busy at 100% and there are processes waiting for it.
It gives the false impression that DRM is still only about defining CPU limitations. This was true until DB version 9i. Now however in fact you can also use DRM to limit excessive I/O (in 11g) or avoid long running queries by canceling them or killing the session altogether, either based on actual runtime or estimates (in 10g already). Check out the documentation in the articles external links.
Regards. --Berny 05:36, 4 June 2010 (UTC)
- I agree but DRM design (as far as I know from an Oracle developer) implies that DRM code is executed ONLY when CPU is at 100% otherwise the code is not executed at all. Maybe it has changed in the latest version (11g) but it was as it before.
- Here's an abstract of my discussion with developer:
- Therefore, the effect of the Database Resource Manager functioning can be noticed only when the system is fully utilized (i.e. CPU utiliation should be at least 100%). In case of multiple CPU's, all the CPUs should be utilized to the maximum.
- Michel Cadot 15:37, 4 June 2010 (UTC)
- I didn't mean that your statement is wrong, it's just incomplete. The added functionality regarding I/O limitation and long running SQL avoidance methods should all be mentioned as well. All of this is part of current DRM releases. I have just recently experimented with the long running SQL avoidance and believe me, that works without any CPU utilization being at 100%. --Berny 20:03, 4 June 2010 (UTC)
- It is just a warning but an important warning as many think that if they configure DMR then all limits are enforced any time and are surprised that some that are in a group limited to say 30% of the CPU can use it at 70%.
- As it the page is in incomplete state and should be improved to precise the limits that can be set and so on.
- Michel Cadot 06:46, 5 June 2010 (UTC)
- Alright, agreed. Maybe this discussion should rather be moved to the discussion page of the article now. What do you think? --Berny 12:57, 6 June 2010 (UTC)
- I think the page should first be completed, then the discussion will have no subject.
- Michel Cadot 13:47, 6 June 2010 (UTC)
Correction to @ and @@ not accepted?
I had attempted to correct a misconception about the usage of @@ for SQL Plus FAQ under the topic of "What is the difference between @ and @@?" It is wrongly believed that @@ used relative directory to script that calls it but actually it is relative the directory from which SQL Plus is launched. Since the launching dir is often be same dir as which the script is contained, this difference is not noticed. Anyway, I made these changes in April 15 but they were rolled back for some reason.
- I rolled back the modification as it is actually relative to the directory of the current script.
- Here's an example:
C:\TEMP>set SQLPATH= C:\TEMP>dir *.sql Le volume dans le lecteur C s'appelle System Le numéro de série du volume est F0C9-0113 Répertoire de C:\TEMP Fichier introuvable C:\TEMP>type c:\t1.sql prompt here I am in t1 prompt now I call t2 @@t2 prompt ending t1 C:\TEMP>type c:\t2.sql prompt this is t2 C:\TEMP>sqlplus michel/michel SQL*Plus: Release 10.2.0.4.0 - Production on Ven. Juil. 9 20:08:10 2010 Copyright (c) 1982, 2007, Oracle. All Rights Reserved. Connected to: Oracle Database 10g Enterprise Edition Release 10.2.0.4.0 - Production With the Partitioning, Oracle Label Security, OLAP, Data Mining and Real Application Testing options SQL> @c:\t1 here I am in t1 now I call t2 this is t2 ending t1
- Michel Cadot 18:14, 9 July 2010 (UTC)
I had removed my original comment because I just retested with a new version and thought it was working:
SQL*Plus: Release 126.96.36.199.0 - Production on Fri Jul 9 11:45:19 2010 Copyright (c) 1982, 2008, Oracle. All rights reserved.
However on re-retesting I did discover some additional strange behavior:
Given a directory structure:
DirA/ master.sql sub.sql DirB/ sub_b.sql
If I start SQL Plus in some other dir than DirA and then use an absolute path to invoke like @DirA/master.sql, then within the master.sql it can reference @@sub.sql but not @@DirB/sub_b.sql.
That is when SQL Plus not launched from same directory as @ caller, it seems to be able to use @@ to reference files within the same directory as the @ caller but not subdirectories.
Chris (user Cdpr)
- Yes this is what I saw too. Michel Cadot 19:30, 9 July 2010 (UTC)
Care if I add in the wiki that the "@@ does not reference scripts within subdirectories relative to the calling script"? That would have saved me some time to know in advance. --Cdpr 20:53, 9 July 2010 (UTC)
- It's OK for me, do it if you want. I should say that "can not reference scripts within directories relative to the calling script" or "can not reference scripts using path relative to the calling script" as you can't call script in any directory (not only subdirectory ones) using relative path; in my example if I replace "t2.sql" by ".\t2.sql" I get an error "unable to open file". Michel Cadot 05:58, 10 July 2010 (UTC)
Okay. Made the change. Feel free to modify. Thanks, --Cdpr 16:23, 12 July 2010 (UTC)
Removal of QueryAdvisor from Freeware
you removed my comment related QueryAdvisor software from Freeware. You can get in addition to a trial license (I did not mention as this is time restricted and you added in discussion a remark about shareware) the free public and the free registered license. You will be able without any restrictions to do the mentioned analysis of tracefiles. The commercial licenses offer additional functionality for testing. Therefor I would like to ask if you could think about your decision again.
Kind regards Frank
- OK I put it back but you can't then add it to Commercial Products page.
- Our policy is to have a product at most in one of the Freeware, Shareware or Commercial Product pages.
- Michel Cadot 19:45, 4 April 2011 (UTC)
Regarding your options page for Oracle
Going forward Audit Vault will be known as Audit Valut and Database Firewall. Yes it is a mouth full, but they are now offered together as an option. Not a big deal and probably ok to leave it as it is as most who are looking for one or the other will find them on a search. But to be precise that is the direction they are headed.
incorrect pricing edit on Statspack page
There is a reference on the
stating my product, Stori is $500/db. This is incorrect because it implies it is $500 per database or Oracle instance. The pricing is $500 per Oracle Database ID (DBID). This means that clones for testing, development, and QA are automatically included in the $500. This single feature makes the pricing extremely attractive to many Statspack users.
Please. Would you either remove the pricing references on the page or change the Stori pricing from $500/db to $500/DBID.
I wouldn't mind so much if the "spviewer" pricing wasn't shown. The way the page reads now, it implies an apples-to-apples comparison, which it is not.
Also, if you think I am abusing my user privileges by changing the page, I am sorry. I don't mean to abuse my privileges. I occasionally check pages where I and my company OraPub are mentioned, just in case something like this occurs.
Thank you for OraFaq. It is a great resource and I mention it in my classes and when I speak to DBAs.
- Requested change (db->DBID) done.
- Many people search only free products, so we added the prices. Also it would not be fair to display in a same way products with different prices: the (short) description shows the differences in features, the price is another indication (of secondary importance, between parentheses).
- For the same reason, products are displayed in the order they come; of course, each one wants his to be the first one in the list so an order had to be chosen, we thought it was the simplest fair way, it could be per price.
- Michel Cadot (talk) 07:44, 19 December 2013 (CET)
hi Michel, please note that Intel Hyperthreading is not considered when counting processor licenses. Actual physical cores are what count. Every two Intel cpu cores count towards one "processor" license, because the .5 is the core factor for Intel. It looks like you still reverted my edit to the Oracle licensing wiki page. Please let me know why you disagree.