<?xml version="1.0" encoding="UTF-8"?>
<rss xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:taxo="http://purl.org/rss/1.0/modules/taxonomy/" version="2.0">
  <channel>
    <title>topic Performance EGuide queries on AIX in SAS Enterprise Guide</title>
    <link>https://communities.sas.com/t5/SAS-Enterprise-Guide/Performance-EGuide-queries-on-AIX/m-p/923#M255</link>
    <description>Hi,&lt;BR /&gt;
&lt;BR /&gt;
We have a situation where multiple users use Enterprise Guide (v3 with hotfix 3 installed) to query&lt;BR /&gt;
star schema's in SAS (v8.2 TS2M0 on AIX)&lt;BR /&gt;
Some queries are resolved by sort-merge, others by index joins.&lt;BR /&gt;
&lt;BR /&gt;
We are running into performance problems with the index joins.&lt;BR /&gt;
SAS data set pages are usually requested from disk multiple times&lt;BR /&gt;
because the index column value is spread over the data set pages&lt;BR /&gt;
(unless it is sorted on the index column values, but that's not always&lt;BR /&gt;
the case).&lt;BR /&gt;
For instance, data set page 1 may contain dates 01jan2000 and&lt;BR /&gt;
02jan2000.&lt;BR /&gt;
When all observations for january 2000 are queried, page 1 will be&lt;BR /&gt;
fetched twice: once for 01jan2000 and once for 02jan2000. This works&lt;BR /&gt;
fine if the system load is low and the data set pages remain in the&lt;BR /&gt;
system file cache between the two reads. However, when the system is&lt;BR /&gt;
busy, these pages are constanly swapped out of the file cache, and have&lt;BR /&gt;
to be fetched from disk again.&lt;BR /&gt;
This thrashing causes performance to fluctuate heavily.&lt;BR /&gt;
When the system load is low, we get a response time of 14 seconds on&lt;BR /&gt;
such a query.&lt;BR /&gt;
When the system load is high, we get a response time of 14 minutes on&lt;BR /&gt;
such a query.&lt;BR /&gt;
Users find this variation very hard to work with.&lt;BR /&gt;
&lt;BR /&gt;
We assigned a file cache on the AIX machine of about 8GB, but SAS data&lt;BR /&gt;
set sizes are often larger than that. When any user submits a query&lt;BR /&gt;
that performs a table scan on such a table, the complete file cache is&lt;BR /&gt;
overwritten by the newly fetched data set pages, even though they are&lt;BR /&gt;
only used once for this table scan. I would rather have not used any&lt;BR /&gt;
cache for the table scan and have left the pages in the cache for use&lt;BR /&gt;
by the index joins.&lt;BR /&gt;
&lt;BR /&gt;
Has anybody else run into this, and found a solution?&lt;BR /&gt;
I would like AIX to use file cache only for index joins and not for&lt;BR /&gt;
table scans.&lt;BR /&gt;
I'm pretty sure that can't be done, but maybe there are alternatives.&lt;BR /&gt;
I could imagine a smarter file caching algorithm that keeps pages in&lt;BR /&gt;
cache depending on their usage.&lt;BR /&gt;
&lt;BR /&gt;
Any suggestions are appreciated.&lt;BR /&gt;
kind regards, &lt;BR /&gt;
Resa Drijsen</description>
    <pubDate>Thu, 29 Jun 2006 10:51:32 GMT</pubDate>
    <dc:creator>Resa</dc:creator>
    <dc:date>2006-06-29T10:51:32Z</dc:date>
    <item>
      <title>Performance EGuide queries on AIX</title>
      <link>https://communities.sas.com/t5/SAS-Enterprise-Guide/Performance-EGuide-queries-on-AIX/m-p/923#M255</link>
      <description>Hi,&lt;BR /&gt;
&lt;BR /&gt;
We have a situation where multiple users use Enterprise Guide (v3 with hotfix 3 installed) to query&lt;BR /&gt;
star schema's in SAS (v8.2 TS2M0 on AIX)&lt;BR /&gt;
Some queries are resolved by sort-merge, others by index joins.&lt;BR /&gt;
&lt;BR /&gt;
We are running into performance problems with the index joins.&lt;BR /&gt;
SAS data set pages are usually requested from disk multiple times&lt;BR /&gt;
because the index column value is spread over the data set pages&lt;BR /&gt;
(unless it is sorted on the index column values, but that's not always&lt;BR /&gt;
the case).&lt;BR /&gt;
For instance, data set page 1 may contain dates 01jan2000 and&lt;BR /&gt;
02jan2000.&lt;BR /&gt;
When all observations for january 2000 are queried, page 1 will be&lt;BR /&gt;
fetched twice: once for 01jan2000 and once for 02jan2000. This works&lt;BR /&gt;
fine if the system load is low and the data set pages remain in the&lt;BR /&gt;
system file cache between the two reads. However, when the system is&lt;BR /&gt;
busy, these pages are constanly swapped out of the file cache, and have&lt;BR /&gt;
to be fetched from disk again.&lt;BR /&gt;
This thrashing causes performance to fluctuate heavily.&lt;BR /&gt;
When the system load is low, we get a response time of 14 seconds on&lt;BR /&gt;
such a query.&lt;BR /&gt;
When the system load is high, we get a response time of 14 minutes on&lt;BR /&gt;
such a query.&lt;BR /&gt;
Users find this variation very hard to work with.&lt;BR /&gt;
&lt;BR /&gt;
We assigned a file cache on the AIX machine of about 8GB, but SAS data&lt;BR /&gt;
set sizes are often larger than that. When any user submits a query&lt;BR /&gt;
that performs a table scan on such a table, the complete file cache is&lt;BR /&gt;
overwritten by the newly fetched data set pages, even though they are&lt;BR /&gt;
only used once for this table scan. I would rather have not used any&lt;BR /&gt;
cache for the table scan and have left the pages in the cache for use&lt;BR /&gt;
by the index joins.&lt;BR /&gt;
&lt;BR /&gt;
Has anybody else run into this, and found a solution?&lt;BR /&gt;
I would like AIX to use file cache only for index joins and not for&lt;BR /&gt;
table scans.&lt;BR /&gt;
I'm pretty sure that can't be done, but maybe there are alternatives.&lt;BR /&gt;
I could imagine a smarter file caching algorithm that keeps pages in&lt;BR /&gt;
cache depending on their usage.&lt;BR /&gt;
&lt;BR /&gt;
Any suggestions are appreciated.&lt;BR /&gt;
kind regards, &lt;BR /&gt;
Resa Drijsen</description>
      <pubDate>Thu, 29 Jun 2006 10:51:32 GMT</pubDate>
      <guid>https://communities.sas.com/t5/SAS-Enterprise-Guide/Performance-EGuide-queries-on-AIX/m-p/923#M255</guid>
      <dc:creator>Resa</dc:creator>
      <dc:date>2006-06-29T10:51:32Z</dc:date>
    </item>
    <item>
      <title>Re: Performance EGuide queries on AIX</title>
      <link>https://communities.sas.com/t5/SAS-Enterprise-Guide/Performance-EGuide-queries-on-AIX/m-p/924#M256</link>
      <description>I know: It's a reply to my own topic but I would like to have the topic on top of te list again to challenge anyone on this forum one more time to come up with suggestions.&lt;BR /&gt;
&lt;BR /&gt;
Looking forward to reactions.</description>
      <pubDate>Thu, 06 Jul 2006 07:35:36 GMT</pubDate>
      <guid>https://communities.sas.com/t5/SAS-Enterprise-Guide/Performance-EGuide-queries-on-AIX/m-p/924#M256</guid>
      <dc:creator>Resa</dc:creator>
      <dc:date>2006-07-06T07:35:36Z</dc:date>
    </item>
  </channel>
</rss>

