BOB: Business Objects Board
Not endorsed by or affiliated with SAP

Register | Login 

Sunset Ride 
Sunset Ride 

General Notice: BOB is going to retire...please see details here. All BOB forums have been converted to "read only" as of September 1st.
General Notice: No events within the next 45 days.

Benchmark YOUR hardware and compare the results V1.1
1 members found this topic helpful
Goto page Previous  1, 2, 3, 4, 5, 6, 7, 8, 9, 10  Next
 
Search this topic... | Search DI: Performance and Tuning... | Search Box
Register or Login to Post    Forum Index -> Data Integrator -> DI: Performance and Tuning  Previous TopicPrint TopicNext Topic
Author Message
dnewton
Forum Addict
Forum Addict



Joined: 30 Jan 2004
ASUG Icon
Posts: 3446
Location: Massachusetts, USA


flag
PostPosted: Thu Dec 11, 2008 8:13 am 
Post subject: Re: Benchmark YOUR hardware and compare the results

For the server (hardware) that is hosting your source and target databases... can you provide some more detail behind your SAN? (Number of disk spindles, RAID configuration, etc.)

ETL processing is so disk-intensive -- particularly given DI's preference to push-down as much as possible to the underlying database -- that if things are configured suboptimally, it can make a big difference.
Back to top
sliceofhogan
Forum Member
Forum Member



Joined: 28 May 2008

Posts: 9


flag
PostPosted: Thu Dec 11, 2008 8:29 am 
Post subject: Re: Benchmark YOUR hardware and compare the results

Here are the new results after setting the connections to SQL 2005 icon_redface.gif :
(11.7) 12-11-08 14:06:46 (5104:5164) JOB: Current system configuration : <AEBDBS>.
(11.7) 12-11-08 14:06:46 (5104:5164) JOB: Reading job <069f43f0_bc6f_4434_9b49_8d0d6b2b451c> from the repository; Server version is <11.7.3.3>; Repository version is
<11.7.0.0000>.
(11.7) 12-11-08 14:06:46 (5104:5164) JOB: Current directory of job <069f43f0_bc6f_4434_9b49_8d0d6b2b451c> is <C:\Program Files\Business Objects\Data Integrator 11.7\bin>.
(11.7) 12-11-08 14:06:47 (5104:5164) JOB: Starting job on job server host <AEBBOJS>, port <3500>.
(11.7) 12-11-08 14:06:48 (5104:5164) JOB: Job <Job_Benchmark_MS> of runid <2008121114064751045164> is initiated by user <boserviceaccount>.
(11.7) 12-11-08 14:06:48 (5104:5164) JOB: Processing job <Job_Benchmark_MS>.
(11.7) 12-11-08 14:07:02 (5104:5164) JOB: Optimizing job <Job_Benchmark_MS>.
(11.7) 12-11-08 14:07:02 (5104:5164) JOB: Job <Job_Benchmark_MS> is started.
(11.7) 12-11-08 14:07:03 (5364:5372) DATAFLOW: Process to execute data flow <DF_Benchmark_Setup_MS> is started.
(11.7) 12-11-08 14:07:06 (5364:5372) DATAFLOW: Data flow <DF_Benchmark_Setup_MS> is started.
(11.7) 12-11-08 14:07:07 (5364:5372) DATAFLOW: Cache statistics determined that data flow <DF_Benchmark_Setup_MS> uses <0> caches with a total size of <0> bytes. This is less
than(or equal to) the virtual memory <1609564160> bytes available for caches. Statistics is switching the cache type to IN
MEMORY.
(11.7) 12-11-08 14:07:08 (5364:5372) DATAFLOW: Data flow <DF_Benchmark_Setup_MS> using IN MEMORY Cache.
(11.7) 12-11-08 14:07:11 (5364:5372) DATAFLOW: Data flow <DF_Benchmark_Setup_MS> is completed successfully.
(11.7) 12-11-08 14:07:12 (5364:5372) DATAFLOW: Process to execute data flow <DF_Benchmark_Setup_MS> is completed.
(11.7) 12-11-08 14:07:13 (5340:5308) DATAFLOW: Process to execute data flow <DF_Benchmark_read_MS> is started.
(11.7) 12-11-08 14:07:18 (5340:5308) DATAFLOW: Data flow <DF_Benchmark_read_MS> is started.
(11.7) 12-11-08 14:07:20 (5340:5308) DATAFLOW: Cache statistics determined that data flow <DF_Benchmark_read_MS> uses <0> caches with a total size of <0> bytes. This is less
than(or equal to) the virtual memory <1608515584> bytes available for caches. Statistics is switching the cache type to IN
MEMORY.
(11.7) 12-11-08 14:07:20 (5340:5308) DATAFLOW: Data flow <DF_Benchmark_read_MS> using IN MEMORY Cache.
(11.7) 12-11-08 14:55:43 (5340:5308) DATAFLOW: Data flow <DF_Benchmark_read_MS> is completed successfully.
(11.7) 12-11-08 14:55:43 (5340:5308) DATAFLOW: Process to execute data flow <DF_Benchmark_read_MS> is completed.
(11.7) 12-11-08 14:55:43 (5104:5164) PRINTFN: Time______________________________ 2911 secounds
(11.7) 12-11-08 14:55:44 (4364:4844) DATAFLOW: Process to execute data flow <DF_Benchmark_API_bulkloader_MS> is started.
(11.7) 12-11-08 14:55:44 (4364:4844) DATAFLOW: Data flow <DF_Benchmark_API_bulkloader_MS> is started.
(11.7) 12-11-08 14:55:44 (4364:4844) DATAFLOW: Cache statistics determined that data flow <DF_Benchmark_API_bulkloader_MS> uses <0> caches with a total size of <0> bytes.
This is less than(or equal to) the virtual memory <1609564160> bytes available for caches. Statistics is switching the cache
type to IN MEMORY.
(11.7) 12-11-08 14:55:44 (4364:4844) DATAFLOW: Data flow <DF_Benchmark_API_bulkloader_MS> using IN MEMORY Cache.


Edited: job is still running...... icon_redface.gif stupidly posted the previous results again. The performance is now worse....


Last edited by sliceofhogan on Thu Dec 11, 2008 9:00 am, edited 1 time in total
Back to top
sliceofhogan
Forum Member
Forum Member



Joined: 28 May 2008

Posts: 9


flag
PostPosted: Thu Dec 11, 2008 8:43 am 
Post subject: Re: Benchmark YOUR hardware and compare the results

Dnewton:

Our SAN configuration is Dell/EqualLogic (2x 4,5Tb RAID50, 6x 1Gb connection). Opposed to popular believe these (iSCSI) arrays are normally never the bottleneck, performance is on par or better than fiber options. A Microsoft Dba has checked our configuration and DB design just last week and was impressed by the performance, specificly the throughput (what we need with BO).
For high transaction environments RAID10 and faster disk arrays can be added.

The reason we did the BO benchmark is that one of our flows stopped processing. This flow uses 1,2mi source rows to match our DWH table (9 mi. rows) using a table compare.

As our SQL server (and databases) are responding correctly we suspect a configuration problem in our BO environment.

According to our Administrators our SQL, BODI and BOJS servers are correctly setup with regards to AntiVirus software,
but will check anyway.
Back to top
Werner Daehn
Forum Devotee
Forum Devotee



Joined: 17 Dec 2004

speaker.gif*6
Posts: 10593
Location: Germany


flag
PostPosted: Thu Dec 11, 2008 9:12 am 
Post subject: Re: Benchmark YOUR hardware and compare the results

rightclick all dataflows and change the properties of them. They should be set to Cachetype: In-Memory for this test. And when executing, disable the flag "use collected statistics".

Only then we can compre your numbers.

_________________
Werner Daehn
ex Product Manager Data Services at SAP SE

www.rtdi.io
Back to top
sliceofhogan
Forum Member
Forum Member



Joined: 28 May 2008

Posts: 9


flag
PostPosted: Thu Dec 11, 2008 9:29 am 
Post subject: Re: Benchmark YOUR hardware and compare the results

thx wdaehn,

followed your instructions: all flags off except enable auditing, all flows changed to in-memory, monitor smaple rate 64.000

However this is the current status (job still running):

/DF_Benchmark_read_MS/Query PROCEED 640000 368.334 378.426

Not really, really quick, to use an understatement..
Back to top
sliceofhogan
Forum Member
Forum Member



Joined: 28 May 2008

Posts: 9


flag
PostPosted: Thu Dec 11, 2008 11:07 am 
Post subject: Re: Benchmark YOUR hardware and compare the results

Here are the completed results of the last run:

4464 1428 DATAFLOW 11-12-2008 15:22:43 Data flow <DF_Benchmark_Setup_MS> is started.
4464 1428 DATAFLOW 11-12-2008 15:22:43 Data flow <DF_Benchmark_Setup_MS> using IN MEMORY Cache.
4464 1428 DATAFLOW 11-12-2008 15:22:44 Data flow <DF_Benchmark_Setup_MS> is completed successfully.
4464 1428 DATAFLOW 11-12-2008 15:22:44 Process to execute data flow <DF_Benchmark_Setup_MS> is completed.
6076 2848 DATAFLOW 11-12-2008 15:22:45 Process to execute data flow <DF_Benchmark_read_MS> is started.
6076 2848 DATAFLOW 11-12-2008 15:22:47 Data flow <DF_Benchmark_read_MS> is started.
6076 2848 DATAFLOW 11-12-2008 15:22:47 Data flow <DF_Benchmark_read_MS> using IN MEMORY Cache.
6076 2848 DATAFLOW 11-12-2008 16:29:10 Data flow <DF_Benchmark_read_MS> is completed successfully.
6076 2848 DATAFLOW 11-12-2008 16:29:10 Process to execute data flow <DF_Benchmark_read_MS> is completed.
1932 1712 PRINTFN 11-12-2008 16:29:11 Time______________________________ 3986 secounds
3344 1280 DATAFLOW 11-12-2008 16:29:12 Process to execute data flow <DF_Benchmark_API_bulkloader_MS> is started.
3344 1280 DATAFLOW 11-12-2008 16:29:12 Data flow <DF_Benchmark_API_bulkloader_MS> is started.
3344 1280 DATAFLOW 11-12-2008 16:29:12 Data flow <DF_Benchmark_API_bulkloader_MS> using IN MEMORY Cache.
3344 1280 DATAFLOW 11-12-2008 16:46:52 Data flow <DF_Benchmark_API_bulkloader_MS> is completed successfully.
3344 1280 DATAFLOW 11-12-2008 16:46:52 Process to execute data flow <DF_Benchmark_API_bulkloader_MS> is completed.
1932 1712 PRINTFN 11-12-2008 16:46:52 Time______________________________ 1061 secounds
5664 1336 DATAFLOW 11-12-2008 16:46:53 Process to execute data flow <DF_Benchmark_regular_load_MS> is started.
5664 1336 DATAFLOW 11-12-2008 16:46:53 Data flow <DF_Benchmark_regular_load_MS> is started.
5664 1336 DATAFLOW 11-12-2008 16:46:53 Data flow <DF_Benchmark_regular_load_MS> using IN MEMORY Cache.
5664 1336 DATAFLOW 11-12-2008 16:52:59 Data flow <DF_Benchmark_regular_load_MS> is completed successfully.
5664 1336 DATAFLOW 11-12-2008 16:52:59 Process to execute data flow <DF_Benchmark_regular_load_MS> is completed.
1932 1712 PRINTFN 11-12-2008 16:52:59 Time______________________________ 367 secounds
3340 4144 DATAFLOW 11-12-2008 16:53:00 Process to execute data flow <DF_Benchmark_single_thread_MS> is started.
3340 4144 DATAFLOW 11-12-2008 16:53:00 Data flow <DF_Benchmark_single_thread_MS> is started.
3340 4144 DATAFLOW 11-12-2008 16:53:00 Data flow <DF_Benchmark_single_thread_MS> using IN MEMORY Cache.
3340 4144 DATAFLOW 11-12-2008 16:53:16 Data flow <DF_Benchmark_single_thread_MS> is completed successfully.
3340 4144 DATAFLOW 11-12-2008 16:53:16 Process to execute data flow <DF_Benchmark_single_thread_MS> is completed.
1932 1712 PRINTFN 11-12-2008 16:53:16 Time______________________________ 16 secounds
3780 4204 DATAFLOW 11-12-2008 16:53:17 Process to execute data flow <DF_Benchmark_lookup_DOP1_MS> is started.
3780 4204 DATAFLOW 11-12-2008 16:53:18 Data flow <DF_Benchmark_lookup_DOP1_MS> is started.
3780 4204 DATAFLOW 11-12-2008 16:53:18 Data flow <DF_Benchmark_lookup_DOP1_MS> using IN MEMORY Cache.
3780 4204 DATAFLOW 11-12-2008 16:53:51 Data flow <DF_Benchmark_lookup_DOP1_MS> is completed successfully.
3780 4204 DATAFLOW 11-12-2008 16:53:51 Process to execute data flow <DF_Benchmark_lookup_DOP1_MS> is completed.
1932 1712 PRINTFN 11-12-2008 16:53:51 Time______________________________ 35 secounds
6116 4524 DATAFLOW 11-12-2008 16:53:52 Process to execute data flow <DF_Benchmark_lookup_DOP10_MS> is started.
6116 4524 DATAFLOW 11-12-2008 16:53:53 Data flow <DF_Benchmark_lookup_DOP10_MS> is started.
6116 4524 DATAFLOW 11-12-2008 16:53:53 Data flow <DF_Benchmark_lookup_DOP10_MS> using IN MEMORY Cache.
6116 4524 DATAFLOW 11-12-2008 16:54:14 Data flow <DF_Benchmark_lookup_DOP10_MS> is completed successfully.
6116 4524 DATAFLOW 11-12-2008 16:54:14 Process to execute data flow <DF_Benchmark_lookup_DOP10_MS> is completed.
1932 1712 PRINTFN 11-12-2008 16:54:14 Time______________________________ 22 secounds
1932 1712 JOB 11-12-2008 16:54:14 Job <Job_Benchmark_MS> is completed successfully.

We can use all the help we can get i suppose!
Back to top
PCL
Forum Member
Forum Member



Joined: 25 Jun 2008

Posts: 5
Location: Netherlands


flag
PostPosted: Wed Jan 21, 2009 8:10 am 
Post subject: Re: Benchmark YOUR hardware and compare the results

Run results of Benchmark after installation of BO Data Service 12.1:

Current system configuration : <AEBDBS>.
Reading job <069f43f0_bc6f_4434_9b49_8d0d6b2b451c> from the repository; Server version is <12.1.1.0>; Repository version is
<12.1.0.0000>.
Current directory of job <069f43f0_bc6f_4434_9b49_8d0d6b2b451c> is <C:\Program Files\Business Objects\Data Services\bin>.
Starting job on job server host <AEBBOJS>, port <3500>.
Job <Job_Benchmark_MS> of runid <200901141118124441804> is initiated by user <boserviceaccount>.
Processing job <Job_Benchmark_MS>.
Optimizing job <Job_Benchmark_MS>.
Job <Job_Benchmark_MS> is started.

Process to execute data flow <DF_Benchmark_read_MS> is completed.
Time 300 secounds

Process to execute data flow <DF_Benchmark_API_bulkloader_MS> is completed.
Time 256 secounds

Process to execute data flow <DF_Benchmark_regular_load_MS> is completed.
Time 269 secounds

Process to execute data flow <DF_Benchmark_single_thread_MS> is completed.
Time 12 secounds

Process to execute data flow <DF_Benchmark_lookup_DOP1_MS> is completed.
Time 30 secounds

Process to execute data flow <DF_Benchmark_lookup_DOP10_MS> is completed.
Time 29 secounds

Job <Job_Benchmark_MS> is completed successfully.
Back to top
Werner Daehn
Forum Devotee
Forum Devotee



Joined: 17 Dec 2004

speaker.gif*6
Posts: 10593
Location: Germany


flag
PostPosted: Wed Jan 21, 2009 8:43 am 
Post subject: Re: Benchmark YOUR hardware and compare the results

What hardware etc? The same like above or different customer...


Source Database version e.g. Oracle 10g
Network link between source and Data Integrator e.g. 1GBit Ethernet
Data Integrator server hardware: Type/CPU/Memory/Disk Windows XP 32bit/4 CPU @ 2.5GHz/2GB memory/Raid 7 with 4 disks
Data Integrator version DI 6.5
Network link between Data Integrator and Target e.g. none - DI and target database installed on same hardware
Target server hardware: Type/CPU/Memory/Disk e.g. DI and target database installed on same server
Target Database version e.g. Oracle 10g

_________________
Werner Daehn
ex Product Manager Data Services at SAP SE

www.rtdi.io
Back to top
Werner Daehn
Forum Devotee
Forum Devotee



Joined: 17 Dec 2004

speaker.gif*6
Posts: 10593
Location: Germany


flag
PostPosted: Wed Jan 28, 2009 11:00 am 
Post subject: Re: Benchmark YOUR hardware and compare the results

What is the impact of the DI Version used?

https://wiki.sdn.sap.com:443/wiki/display/BOBJ/Results+with+different+DI+Versions

What would be the impact of the Commit Rate in this benchmark?
https://www.sdn.sap.com/irj/scn/wiki?path=/display/BOBJ/The+impact+of+the+CommitSize
(includes sub chapter for Oracle and SQL Server)


What would be the impact of the number of loaders parameter?
https://wiki.sdn.sap.com:443/wiki/display/BOBJ/The+Impact+of+Number+of+Loaders
(includes sub chapter for Oracle and SQL Server)

_________________
Werner Daehn
ex Product Manager Data Services at SAP SE

www.rtdi.io
Back to top
eepjr24
Forum Enthusiast
Forum Enthusiast



Joined: 16 Sep 2005
ASUG Icon
Posts: 1815
Location: Orlando, FL


flag
PostPosted: Wed Jan 28, 2009 11:11 am 
Post subject: Re: Benchmark YOUR hardware and compare the results

Interesting read, thanks. I need to go and check a couple jobs where we do the loaders per partition and make sure that is really our best bet. Most of the rest falls in line with my anecdotal experiences in 11.7 on Oracle.

- Ernie

_________________
Please Search before you post.
___________________________
Ernie Phelps
Senior Manager
Protiviti
http://www.protiviti.com
Back to top
dnewton
Forum Addict
Forum Addict



Joined: 30 Jan 2004
ASUG Icon
Posts: 3446
Location: Massachusetts, USA


flag
PostPosted: Wed Jan 28, 2009 2:28 pm 
Post subject: Re: Benchmark YOUR hardware and compare the results

I've done similar testing, but using much wider tables than the standard test suite that Werner's published. I think the answer is, it really depends upon the dataflow and target table.

For example, with narrower tables in SQL Server, changing the # of Loaders from 1 to 2 can get you that 30% boost. But with wider target tables, where those tables have a lot of indexes in place, the contention becomes more of an issue and you don't get that 30%.

We've also found that changing the SQL Server bulkloader settings between, say, 5000 rows and 20000 rows, can make a big difference with some tables. I'm not sure why...
Back to top
eepjr24
Forum Enthusiast
Forum Enthusiast



Joined: 16 Sep 2005
ASUG Icon
Posts: 1815
Location: Orlando, FL


flag
PostPosted: Wed Jan 28, 2009 2:34 pm 
Post subject: Re: Benchmark YOUR hardware and compare the results

nods. We have seen our share of "interesting" improvements as well, but almost any system that utilizes an optimizer as complex as either DI's or Oracles is going to run into those type of situations. And tuning will forever be a delicate balance of art and science, in my book. I have spent alot of time trying to code the "art" out of it and all I have ended up doing is getting better at the art part. eek.gif

- Ernie

_________________
Please Search before you post.
___________________________
Ernie Phelps
Senior Manager
Protiviti
http://www.protiviti.com
Back to top
Werner Daehn
Forum Devotee
Forum Devotee



Joined: 17 Dec 2004

speaker.gif*6
Posts: 10593
Location: Germany


flag
PostPosted: Fri Jan 30, 2009 1:15 pm 
Post subject: Re: Benchmark YOUR hardware and compare the results

My last day with the Lab-Server yesnod.gif

I have installed VMWare and inside ran the DI jobserver. The database remained on the physical machine with access to all 8 cores. As the VMWare can only offer one or two CPUs in the guest OS and I needed something to compare with, I did set the CPU affinity to one CPU, two CPUs on different dies and 2 CPUs sharing the L3 cache.

Read: No matter what I do on the host installation with DI, I keep reading within 80secs constant. VMWare with two CPUs is 116secs, with one CPU 182secs. I'd say this is the additional CPU costs of simulating a network.

https://wiki.sdn.sap.com:443/wiki/display/BOBJ/Results

_________________
Werner Daehn
ex Product Manager Data Services at SAP SE

www.rtdi.io
Back to top
gpn01
Principal Member
Principal Member



Joined: 30 Nov 2004

Posts: 224
Location: London, UK


flag
PostPosted: Mon Mar 30, 2009 9:16 am 
Post subject: Re: Benchmark YOUR hardware and compare the results

I keep getting "Remote server not responding" when trying to download the .zip file. Has the file moved?
Back to top
Werner Daehn
Forum Devotee
Forum Devotee



Joined: 17 Dec 2004

speaker.gif*6
Posts: 10593
Location: Germany


flag
PostPosted: Mon Mar 30, 2009 12:07 pm 
Post subject: Re: Benchmark YOUR hardware and compare the results

please use IE to download/view the page. I have raised that problem within SAP.
_________________
Werner Daehn
ex Product Manager Data Services at SAP SE

www.rtdi.io
Back to top
Display posts from previous:   
Register or Login to Post    Forum Index -> Data Integrator -> DI: Performance and Tuning  Previous TopicPrint TopicNext Topic
Page 5 of 10 All times are GMT - 5 Hours
Goto page Previous  1, 2, 3, 4, 5, 6, 7, 8, 9, 10  Next
 
Jump to:  

Index | About | FAQ | RAG | Privacy | Search |  Register |  Login 

Not endorsed by or affiliated with SAP
Powered by phpBB © phpBB Group
Generated in 0.1007 seconds using 17 queries. (SQL 0.0027 Parse 0.0679 Other 0.0301)
CCBot/2.0 (https://commoncrawl.org/faq/)
Hosted by ForumTopics.com | Terms of Service
phpBB Customizations by the phpBBDoctor.com
Shameless plug for MomentsOfLight.com Moments of Light Logo