SAS ANZ ran a forum in Sydney this month, and had a great turn out with over 1,000 attendies. I didn’t manage to make it but heard it was a great event.
SAS ANZ have put up copies of the papers over on the SAS Forum Australia and New Zealand 2008 conference web site, check them out, there is some good info and highlights on whats comming in 9.2!
On a project I am working on we upgraded our SAS licenses from Data Integration Server to Enterprise Data Integration Server.
We were really impressed with the capabilities of Dataflux, and as we were using SAS DI Studio in anger to populate our warehouse, the idea of integrating the Dataflux rules into our ETL processes also appealed.
We had a demo of Dataflux version 8, which looks sexy and has a lot of great features (which you would expect from one of the top 3 data quality tools in the world).
Imagine our disappointment when we got the CD’s to find we had been shipped Dataflux 7.0, not version 8 that was demo’d.
Simple mistake we thought, so onto to our capable SAS account manager.
Well we were wrong, it seems that Dataflux 7.0 is certified (and works) with DI Studio, Dataflux 7.1 and 8.x will be supported in SAS 9.2.
Tricks for young players I suppose., but then again I am not so young anymore…..
I am sure Dataflux 7.0 and DI Studio will do what we need, but still slightly disappointing.
Just spent the week attending the SAS Platform Administrator Fast Trackcourse.
I would highly recomend this course if you have to administer SAS9 Servers or just want a better understanding of how all the SAS 9 stuff fits together.
If you ever have to touch security in SAS 9 then you need to do this course!
Course outline is here: http://support.sas.com/training/us/crs/pafast.html
I will be doing a few posts over the next week to remind me of the things that I picked up and really want to remember.
Mental note to self…..
There is a free tool, called Process Explorer, for Windows servers that provides enhanced monitoring of processes that are running on the servers. Much better than the standard windows taskman or perfmon.
You can downlaod it from here:
http://technet.microsoft.com/en-us/sysinternals/bb896653.aspx
As I mentioned earlier we are trying to do some testing around performance tuning our queries (well I say we but Evan Wilson is doing all the real work!)
So I have been looking for papers that provide some ideas on how to monitor or tune the SAS environment to make things go faster.
I have posted links and abstracts on any papers I found over on our main website under SAS Forum Papers – Performance Tuning
Let me know if you know of any papers I have missed.
Well our developers have been chugging along building our SAS Portlets (and I have been much less efficient in creating the required documentation for them).
I have created the install guides for Tomcat and BEA Weblogic environments, but I don’t have access to a IBM Websphere environment, so if you have said environment and you are willing to swap some documentation and screenshots for a free copy of one of our portlets, send me a note!
Mental note to self, if you want to add a new deployment directory to deploy jobs too, for the scheduler then:
Within SAS Management Console right click on the Schedule Manager folder/tree and the select create new deployment directory.
Details are on page 142 of this, http://support.sas.com/documentation/onlinedoc/91pdf/sasdoc_913/mgmtconsole_ug_9952.pdf
Found an interesting paper by John M. Wildenthal of JPMorgan Chase called An Effective CLEANWORK.SAS® for Windows , outlining a way of automatically deleting work files for any SAS processes that have been terminated abnormally.
Abstract of the paper is:
SAS provides an effective utility for deleting orphaned work directories on Unix boxes. But their offering for Windows
merely deletes everything older than a particular time (http://support.sas.com/techsup/unotes/SN/008/008786.html).
This is unworkable for a server environment. This paper discusses a short program for Windows that parses the root
WORK directory and matches directory names to SAS v9 running processes. Directories that are not matched are
then deleted.
We are currently trying to monitor some SAS queries against our new Star Schemas to see if we can make them go faster.
So I have been doing a bit of research to see whats out there to help us do the monitoring.
In my search I stumbled across this article:
Monitoring SAS® through the Web
ABSTRACT
Monitoring SAS processes and SAS work space in a mutli-user environment, such as UNIX, is key in maintaining optimal performance. This paper will present the SAS Monitor, a tool which captures SAS resources and uses SAS to analyze and display this information via a Web browser. SAS Monitor is useful for SAS/UNIX administrators and informative for end users since the information can be widely distributed through an intranet. Even though SAS Monitor has been tested on Solaris, it can be adapted to other flavors of UNIX. This paper will explore and explain some of the techniques used in capturing and publishing SAS resources on the web.They also provide example code for this at SAS Monitor – Source Code
Unfortunately the code is based upon Unix and we are running windows. So the search continues….
As an aside we are still building the Star Schema’s using SAS Datasets but we plan to test SAS SPDS vs Oracle as a repository for the Star Schemas at the end of the year to see which performs better. If one results in any noticable improvement we may move to it, as it will also solve any file locking issues we currently encounter with SAS Datasets. More on that later.