Shared Memory VS Network : Performance
A question about how much performance is lost connecting to Informix over TCP (onsoctcp) instead of shared memory (onipcshm). Replies varied: one site saw no noticeable difference after moving to a separate DB server, another measured ~20% slower dbimport over TCP. Art Kagel explained the gap is mostly network latency/throughput plus a small protocol overhead, suggested stream pipes for local clients, and noted that offloading clients often frees CPU/memory for the engine, offsetting the loss after retuning. It was also advised to set the FET_BUF_SIZE environment variable (e.g. 32000) before running dbimport/dbexport to speed transfers.
Auto-generated by DrWatson from the posts below — may be imperfect; read the full thread.
Just out of curiosity. Assumings the hardware is all the same and software has a low impact, how much efficiency is lost when going from shared memory access to Informix to network access to Informix. (assuming you have a decent network ) Jonathon Wyza CX & CBORD System Administrator CX Programmer/Analyst Administrative Computing Bethel College (574)-257-3381 AIM: Iamwyza Skype: jonathon.wyza jonathon.wyza@bethelcollege.edu ============================== SLES 10x32 SP2 & IDS 10.00.UC9 " I would love to change the world, but they won't give me the source code." -- Unknown
Hi Jonathon. We at the Claremont Colleges have a separate Informix DB server for our CX system. I have notice no loss of performance with onsoctcp in the sqlhosts (replacing onipcshm). If you need any details with Informix on a separate DB server with CX, please feel free to call me. Jonathan B. Smaby Pomona College phone: (909) 621-8506 email: jonathan.smaby@pomona.edu Semper Paratus! "Always Ready!" -----Original Message----- From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of Wyza, Jonathon Sent: Thursday, April 22, 2010 9:25 PM To: ids@iiug.org Subject: Shared Memory VS Network : Performance [19799] Just out of curiosity. Assumings the hardware is all the same and software has a low impact, how much efficiency is lost when going from shared memory access to Informix to network access to Informix. (assuming you have a decent network ) Jonathon Wyza CX & CBORD System Administrator CX Programmer/Analyst Administrative Computing Bethel College (574)-257-3381 AIM: Iamwyza Skype: jonathon.wyza jonathon.wyza@bethelcollege.edu ============================== SLES 10x32 SP2 & IDS 10.00.UC9 " I would love to change the world, but they won't give me the source code." -- Unknown ******************************************************************************* Forum Note: Use "Reply" to post a response in the discussion forum. No virus found in this incoming message. Checked by AVG - www.avg.com Version: 9.0.814 / Virus Database: 271.1.1/2828 - Release Date: 04/22/10 11:31:00 ------------------------------------------------------------- This message has been scanned by Postini anti-virus software.
Hi Jonathon:
We have notice significant performance improvement (say 20% )while doing
dbimport using share memory over tcp.
Paul
-----Original Message-----
From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of Jonathan
Smaby
Sent: Friday, 23 April 2010 2:51 PM
To: ids@iiug.org
Subject: RE: Shared Memory VS Network : Performance [19800]
Hi Jonathon.
We at the Claremont Colleges have a separate Informix DB server for our CX
system. I have notice no loss of performance with onsoctcp in the sqlhosts
(replacing onipcshm).
If you need any details with Informix on a separate DB server with CX, please
feel free to call me.
Jonathan B. Smaby
Pomona College
phone: (909) 621-8506
email: jonathan.smaby@pomona.edu
Semper Paratus! "Always Ready!"
-----Original Message-----
From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of Wyza,
Jonathon
Sent: Thursday, April 22, 2010 9:25 PM
To: ids@iiug.org
Subject: Shared Memory VS Network : Performance [19799]
Just out of curiosity. Assumings the hardware is all the same and software has
a low impact, how much efficiency is lost when going from shared memory access
to Informix to network access to Informix. (assuming you have a decent network
)
Jonathon Wyza
CX & CBORD System Administrator
CX Programmer/Analyst
Administrative Computing
Bethel College
(574)-257-3381
AIM: Iamwyza
Skype: jonathon.wyza
jonathon.wyza@bethelcollege.edu
==============================
SLES 10x32 SP2 & IDS 10.00.UC9
" I would love to change the world, but they won't give me the source code."
-- Unknown
*******************************************************************************
Forum Note: Use "Reply" to post a response in the discussion forum.
No virus found in this incoming message.
Checked by AVG - www.avg.com
Version: 9.0.814 / Virus Database: 271.1.1/2828 - Release Date: 04/22/10
11:31:00
-------------------------------------------------------------
This message has been scanned by Postini anti-virus software.
*******************************************************************************
Forum Note: Use "Reply" to post a response in the discussion forum.
It's mostly network latency and transmission rate that affect a difference between shared memory connections and networked connections. However, there is a small overhead component which can be seen when comparing a local client on the same machine using shared memory versus using a local loopback network connection. It's not very big though. If the clients are local you can also use stream pipe connections which are between shared memory and networked connections in speed, but if you are considering moving clients off the server onto another machine networked connections are the way to go with the fastest, least shared network you can manage. Usually the ability to have the server take better advantage of the hardware that it's running on, since it is no longer competing with clients for resources, outweighs any loss to the network and you will notice no loss of throughput or responsiveness. You'll just have to retune the engine to use more memory and cpu resources that you have now freed up. Art Art S. Kagel Advanced DataTools (www.advancedatatools.com) IIUG Board of Directors (art@iiug.org) See you at the 2010 IIUG Informix Conference April 25-28, 2010 Overland Park (Kansas City), KS www.iiug.org/conf Disclaimer: Please keep in mind that my own opinions are my own opinions and do not reflect on my employer, Advanced DataTools, the IIUG, nor any other organization with which I am associated either explicitly, implicitly, or by inference. Neither do those opinions reflect those of other individuals affiliated with any entity with which I am affiliated nor those of the entities themselves. On Fri, Apr 23, 2010 at 12:25 AM, Wyza, Jonathon <wyzaj@bethelcollege.edu>wrote: > Just out of curiosity. Assumings the hardware is all the same and software > has > a low impact, how much efficiency is lost when going from shared memory > access > to Informix to network access to Informix. (assuming you have a decent > network > ) > > Jonathon Wyza > CX & CBORD System Administrator > CX Programmer/Analyst > Administrative Computing > Bethel College > (574)-257-3381 > AIM: Iamwyza > Skype: jonathon.wyza > jonathon.wyza@bethelcollege.edu > ============================== > SLES 10x32 SP2 & IDS 10.00.UC9 > > " I would love to change the world, but they won't give me the source > code." > -- Unknown > > > > ******************************************************************************* > Forum Note: Use "Reply" to post a response in the discussion forum. > > --00504502c07876bc080484e53a04
You should look at setting FET_BUF_SIZE to 32000 before
starting dbimport/dbexport.
John F. Miller III
STSM, Support Architect
miller3@us.ibm.com
503-578-5645
IBM Informix Dynamic Server (IDS)
ids-bounces@iiug.org wrote on 04/22/2010 10:19:07 PM:
> [image removed]
>
> RE: Shared Memory VS Network : Performance [19801]
>
> Chin, Paul (TSG-OSDO)
>
> to:
>
> ids
>
> 04/22/2010 10:19 PM
>
> Sent by:
>
> ids-bounces@iiug.org
>
> Please respond to ids
>
> Hi Jonathon:
>
> We have notice significant performance improvement (say 20% )while doing
> dbimport using share memory over tcp.
>
> Paul
>
> -----Original Message-----
> From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf
> Of Jonathan
> Smaby
> Sent: Friday, 23 April 2010 2:51 PM
> To: ids@iiug.org
> Subject: RE: Shared Memory VS Network : Performance [19800]
>
> Hi Jonathon.
>
> We at the Claremont Colleges have a separate Informix DB server for our
CX
> system. I have notice no loss of performance with onsoctcp in the
sqlhosts
> (replacing onipcshm).
>
> If you need any details with Informix on a separate DB server with CX,
please
> feel free to call me.
>
> Jonathan B. Smaby
> Pomona College
> phone: (909) 621-8506
> email: jonathan.smaby@pomona.edu
> Semper Paratus! "Always Ready!"
>
> -----Original Message-----
> From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of
Wyza,
> Jonathon
> Sent: Thursday, April 22, 2010 9:25 PM
> To: ids@iiug.org
> Subject: Shared Memory VS Network : Performance [19799]
>
> Just out of curiosity. Assumings the hardware is all the same and
> software has
> a low impact, how much efficiency is lost when going from shared
> memory access
> to Informix to network access to Informix. (assuming you have a
> decent network
> )
>
> Jonathon Wyza
> CX & CBORD System Administrator
> CX Programmer/Analyst
> Administrative Computing
> Bethel College
> (574)-257-3381
> AIM: Iamwyza
> Skype: jonathon.wyza
> jonathon.wyza@bethelcollege.edu
> ==============================
> SLES 10x32 SP2 & IDS 10.00.UC9
>
> " I would love to change the world, but they won't give me the source
code."
> -- Unknown
>
>
>
*******************************************************************************
> Forum Note: Use "Reply" to post a response in the discussion forum.
>
> No virus found in this incoming message.
> Checked by AVG - www.avg.com
> Version: 9.0.814 / Virus Database: 271.1.1/2828 - Release Date: 04/22/10
> 11:31:00
>
> -------------------------------------------------------------
> This message has been scanned by Postini anti-virus software.
>
>
>
*******************************************************************************
> Forum Note: Use "Reply" to post a response in the discussion forum.
>
>
>
*******************************************************************************
> Forum Note: Use "Reply" to post a response in the discussion forum.
>
John,
That's great! I use dbimport/export for different IDS platforms (HP-UX ->
Linux). Should that FET_BUG_SIZE be in the ONCONFIG file? Or Environment
variable?
Thanks again!
Jonathan B. Smaby
Pomona College
phone: (909) 621-8506
email: jonathan.smaby@pomona.edu
Semper Paratus! "Always Ready!"
-----Original Message-----
From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of John
Miller iii
Sent: Friday, April 23, 2010 10:08 AM
To: ids@iiug.org
Subject: RE: Shared Memory VS Network : Performance [19823]
You should look at setting FET_BUF_SIZE to 32000 before
starting dbimport/dbexport.
John F. Miller III
STSM, Support Architect
miller3@us.ibm.com
503-578-5645
IBM Informix Dynamic Server (IDS)
ids-bounces@iiug.org wrote on 04/22/2010 10:19:07 PM:
> [image removed]
>
> RE: Shared Memory VS Network : Performance [19801]
>
> Chin, Paul (TSG-OSDO)
>
> to:
>
> ids
>
> 04/22/2010 10:19 PM
>
> Sent by:
>
> ids-bounces@iiug.org
>
> Please respond to ids
>
> Hi Jonathon:
>
> We have notice significant performance improvement (say 20% )while doing
> dbimport using share memory over tcp.
>
> Paul
>
> -----Original Message-----
> From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf
> Of Jonathan
> Smaby
> Sent: Friday, 23 April 2010 2:51 PM
> To: ids@iiug.org
> Subject: RE: Shared Memory VS Network : Performance [19800]
>
> Hi Jonathon.
>
> We at the Claremont Colleges have a separate Informix DB server for our
CX
> system. I have notice no loss of performance with onsoctcp in the
sqlhosts
> (replacing onipcshm).
>
> If you need any details with Informix on a separate DB server with CX,
please
> feel free to call me.
>
> Jonathan B. Smaby
> Pomona College
> phone: (909) 621-8506
> email: jonathan.smaby@pomona.edu
> Semper Paratus! "Always Ready!"
>
> -----Original Message-----
> From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of
Wyza,
> Jonathon
> Sent: Thursday, April 22, 2010 9:25 PM
> To: ids@iiug.org
> Subject: Shared Memory VS Network : Performance [19799]
>
> Just out of curiosity. Assumings the hardware is all the same and
> software has
> a low impact, how much efficiency is lost when going from shared
> memory access
> to Informix to network access to Informix. (assuming you have a
> decent network
> )
>
> Jonathon Wyza
> CX & CBORD System Administrator
> CX Programmer/Analyst
> Administrative Computing
> Bethel College
> (574)-257-3381
> AIM: Iamwyza
> Skype: jonathon.wyza
> jonathon.wyza@bethelcollege.edu
> ==============================
> SLES 10x32 SP2 & IDS 10.00.UC9
>
> " I would love to change the world, but they won't give me the source
code."
> -- Unknown
>
>
>
*******************************************************************************
> Forum Note: Use "Reply" to post a response in the discussion forum.
>
> No virus found in this incoming message.
> Checked by AVG - www.avg.com
> Version: 9.0.814 / Virus Database: 271.1.1/2828 - Release Date: 04/22/10
> 11:31:00
>
> -------------------------------------------------------------
> This message has been scanned by Postini anti-virus software.
>
>
>
*******************************************************************************
> Forum Note: Use "Reply" to post a response in the discussion forum.
>
>
>
*******************************************************************************
> Forum Note: Use "Reply" to post a response in the discussion forum.
>
*******************************************************************************
Forum Note: Use "Reply" to post a response in the discussion forum.
No virus found in this incoming message.
Checked by AVG - www.avg.com
Version: 9.0.814 / Virus Database: 271.1.1/2830 - Release Date: 04/22/10
23:31:00
-------------------------------------------------------------
This message has been scanned by Postini anti-virus software.
Environment variable.
http://publib.boulder.ibm.com/infocenter/idshelp/v115/index.jsp?topic=/com.ibm.s
qlr.doc/ids_sqr_235.htm&resultof=%22fet_buf_size%22
On Fri, Apr 23, 2010 at 1:21 PM, Jonathan Smaby
<Jonathan.Smaby@pomona.edu>wrote:
> John,
>
> That's great! I use dbimport/export for different IDS platforms (HP-UX ->
> Linux). Should that FET_BUG_SIZE be in the ONCONFIG file? Or Environment
> variable?
>
> Thanks again!
>
> Jonathan B. Smaby
> Pomona College
> phone: (909) 621-8506
> email: jonathan.smaby@pomona.edu
> Semper Paratus! "Always Ready!"
>
> -----Original Message-----
> From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of John
> Miller iii
> Sent: Friday, April 23, 2010 10:08 AM
> To: ids@iiug.org
> Subject: RE: Shared Memory VS Network : Performance [19823]
>
> You should look at setting FET_BUF_SIZE to 32000 before
> starting dbimport/dbexport.
>
> John F. Miller III
> STSM, Support Architect
> miller3@us.ibm.com
> 503-578-5645
> IBM Informix Dynamic Server (IDS)
>
> ids-bounces@iiug.org wrote on 04/22/2010 10:19:07 PM:
>
> > [image removed]
> >
> > RE: Shared Memory VS Network : Performance [19801]
> >
> > Chin, Paul (TSG-OSDO)
> >
> > to:
> >
> > ids
> >
> > 04/22/2010 10:19 PM
> >
> > Sent by:
> >
> > ids-bounces@iiug.org
> >
> > Please respond to ids
> >
> > Hi Jonathon:
> >
> > We have notice significant performance improvement (say 20% )while doing
> > dbimport using share memory over tcp.
> >
> > Paul
> >
> > -----Original Message-----
> > From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf
> > Of Jonathan
> > Smaby
> > Sent: Friday, 23 April 2010 2:51 PM
> > To: ids@iiug.org
> > Subject: RE: Shared Memory VS Network : Performance [19800]
> >
> > Hi Jonathon.
> >
> > We at the Claremont Colleges have a separate Informix DB server for our
> CX
> > system. I have notice no loss of performance with onsoctcp in the
> sqlhosts
> > (replacing onipcshm).
> >
> > If you need any details with Informix on a separate DB server with CX,
> please
> > feel free to call me.
> >
> > Jonathan B. Smaby
> > Pomona College
> > phone: (909) 621-8506
> > email: jonathan.smaby@pomona.edu
> > Semper Paratus! "Always Ready!"
> >
> > -----Original Message-----
> > From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of
> Wyza,
> > Jonathon
> > Sent: Thursday, April 22, 2010 9:25 PM
> > To: ids@iiug.org
> > Subject: Shared Memory VS Network : Performance [19799]
> >
> > Just out of curiosity. Assumings the hardware is all the same and
> > software has
> > a low impact, how much efficiency is lost when going from shared
> > memory access
> > to Informix to network access to Informix. (assuming you have a
> > decent network
> > )
> >
> > Jonathon Wyza
> > CX & CBORD System Administrator
> > CX Programmer/Analyst
> > Administrative Computing
> > Bethel College
> > (574)-257-3381
> > AIM: Iamwyza
> > Skype: jonathon.wyza
> > jonathon.wyza@bethelcollege.edu
> > ==============================
> > SLES 10x32 SP2 & IDS 10.00.UC9
> >
> > " I would love to change the world, but they won't give me the source
> code."
> > -- Unknown
> >
> >
> >
>
>
>
>
*******************************************************************************
>
> > Forum Note: Use "Reply" to post a response in the discussion forum.
> >
> > No virus found in this incoming message.
> > Checked by AVG - www.avg.com
> > Version: 9.0.814 / Virus Database: 271.1.1/2828 - Release Date: 04/22/10
> > 11:31:00
> >
> > -------------------------------------------------------------
> > This message has been scanned by Postini anti-virus software.
> >
> >
> >
>
>
>
>
*******************************************************************************
>
> > Forum Note: Use "Reply" to post a response in the discussion forum.
> >
> >
> >
>
>
>
>
*******************************************************************************
>
> > Forum Note: Use "Reply" to post a response in the discussion forum.
> >
>
>
>
>
*******************************************************************************
> Forum Note: Use "Reply" to post a response in the discussion forum.
>
> No virus found in this incoming message.
> Checked by AVG - www.avg.com
> Version: 9.0.814 / Virus Database: 271.1.1/2830 - Release Date: 04/22/10
> 23:31:00
>
> -------------------------------------------------------------
> This message has been scanned by Postini anti-virus software.
>
>
>
>
*******************************************************************************
> Forum Note: Use "Reply" to post a response in the discussion forum.
>
>
--001636cd73dd8cbb0e0484eb2e98