Date: Tue, 6 Mar 2012 13:34:27 -0600
Reply-To: Warren Schlechte <Warren.Schlechte@TPWD.STATE.TX.US>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: Warren Schlechte <Warren.Schlechte@TPWD.STATE.TX.US>
Subject: Re: Switching from Local PC version of SAS to Server Based
version Pros and Cons
In-Reply-To: <201203061919.q26FU3LP023765@waikiki.cc.uga.edu>
Content-Type: text/plain; charset="us-ascii"
The server is at a remote location that we access via our DSL.
Hence my statement to the effect you are only as fast as your slowest
part. It is my opinion our bottleneck is our bandwidth.
Warren Schlechte
Learn how you can help Texas State Parks
-----Original Message-----
From: Ben Powell [mailto:bpowell555@GMAIL.COM]
Sent: Tuesday, March 06, 2012 1:20 PM
To: SAS-L@LISTSERV.UGA.EDU; Warren Schlechte
Subject: Re: Switching from Local PC version of SAS to Server Based
version Pros and Cons
Hi Warren,
Can you confirm the location of the server, is it at a remote location?
Compute time
seems to be comparable but you've got some delay in transmission of the
resulting
output,
Regards
On Tue, 6 Mar 2012 09:51:06 -0600, Warren Schlechte
<Warren.Schlechte@TPWD.STATE.TX.US> wrote:
>I decided to run a very simple test to show the differences we
>experience comparing a local install to a server install. I am hoping
>that others might suggest where we should look for improvements in
>efficiency.
>
>Here's the code I submitted. The code is local and the report displays
>locally.
>
>Data test;
>Do I = 1 to 100000;
>Output;
>End;
>Run;
>
>Proc summary printall;
>Var I;
>Run;
>
>I pulled the real and cpu times out of the log file. I ran the code
>multiple times and took the averages for the tables below. The way to
>read the table below is to note that when I run it locally it takes
less
>than 1 sec from submission to displayed output, even with online video
>streaming. However, it can take the same code almost 30 seconds if I
>send it to the server and have high bandwidth use. It takes 4-5
seconds
>with low to moderate bandwidth use. Where I work, we are on DSL.
>Today, our speed is averaging 1.0 Mbps.
>
>Local (With Video Streaming)
> Data Summary Ttl
>Real 0.01 0.02 1
>CPU 0.01 0.02
>
>Server-Based
>Without Video Streaming With Video Streaming
>
> Data Summary Ttl Data Summary Ttl
>
>Real 0.05 0.16 4.33 0.07 0.15 26.67
>
>CPU 0.03 0.11 0.04 0.15
>
>
>Warren Schlechte
>Learn how you can help Texas State Parks
>
>-----Original Message-----
>From: Ben Powell [mailto:bpowell555@GMAIL.COM]
>Sent: Tuesday, March 06, 2012 3:03 AM
>Subject: Re: Switching from Local PC version of SAS to Server Based
>version Pros and Cons
>
>As the previous reply suggested local bandwidth would only be an issue
>here
>if you were using large local datasets which does not appear to be the
>case.
>As such I would rule local bandwidth out as a potential cause of your
>slow
>server performance.
>
>Even if you do have a bandwidth constraint that would not impact on the
>the
>server being able to do its job. So, you would need to get a handle on
>the
>typical server bottlenecks which might be memory, CPU, I/O, or
>fileserver/database connection performance, if the data is not all on
>the
>SAS server.
>
>If using a Win box, taskmgr; linux I like htop; tea-leaves if you're
>using
>MF. If you're using a Mac best pay a visit to your local genius bar and
>have
>a latte,
>
>Regards
| 
