what to do to reduce COB time?

[bdf]

greetings

Cob I work for take to long, can anybody suggest what to do?
I already distributed and resize some of jobs like
FBNK.STMT.ENTRY,FBNK.CATEG.ENTRY,FBNK.RE.CONSOL.SPEC.ENTRY, and it
seems doesn't work with the time i have on my Cob.

Can anybody suggest what to do?

Regards & thanks

Yudi

jsh t24it ~ -->jdiag
jdiag - jBASE diagnostic '$Revision: 1.13 $'

System Information
==================

System : HP-UX shintapd B.11.23.U ia64
UNIX User : t24it (uid 120, euid 120)
Tty name : /dev/pts/tm
Time : Thu Feb 21 09:31:33 2008

Environment
===========

JBCPORTNO : Not Set
JBCRELEASEDIR : '/usr/jbc'
JBCGLOBALDIR : '/usr/jbc'
WARNING: JBCDATADIR is not set, Default '/usr/jbc/jbase_data'
WARNING: JBCDATADIR is subdirectory of JBCGLOBALDIR
HOME : '/database/t24it/bnk/bnk.run'
JEDIFILEPATH : '/database/t24it/bnk/bnk.run'
JEDIFILENAME_MD : '/database/t24it/bnk/bnk.run/VOC'
JEDIFILENAME_SYSTEM : '/usr/jbc/src/SYSTEM'
SYSTEM File is (DICT) : '/usr/jbc/src/SYSTEM]D'
RELEASE Information : Major 4.1 , Minor 5.17 , Patch 5690
(Change 52756)
Spooler dir (JBCSPOOLERDIR) : '/usr/jbc/jbase_data/jspooler'
JBCEMULATE : 'prime'
Object path (JBCOBJECTLIST) : '/database/t24it/bnk/bnk.run/lib:/
database/t24it/b
nk/bnk.run/globuslib'
jBASE Compiler Run-time : '/usr/jbc/config/system.properties'
Program dir (JBCDEV_BIN) : '/database/t24it/bnk/bnk.run/bin'
Subroutine dir (JBCDEV_LIB) : '/database/t24it/bnk/bnk.run/lib'

[Jim Idle]

Switch from HPUX, especially on the Itanic chips? ;-)

Seriously though, as far as I am aware, the COB is essentially disk
bound. So:

1) Have you tuned your HPUX cache so that it dedicates more RAM to disk
IO caching? This is a common issue on HPUX systems;
2) Add more memory
3) Upgrade the disk IO system;
4) Call IBM ;-)

Jim

[Andrew Trusov]

Hi!
To improve your COB performance you can use
.NTX option in PGM.FILE  3.ADDITIONAL.INFO for reporting jobs.
For details  see

http://groups.google.com/group/jBASE/browse_frm/thread/aa0f45c4ebe209d1

2008/2/21, bdf <yd8...@gmail.com>:

--
А писал все это Барисыч - Google'оид со стажем.

[bdf]

hi Andrew,
from my perspective i think what you suggest isn't quite relevant, but
many thanks
any other suggestion guys?

regards
Yudi

[Mike Preece]

This is a *huge* question. There are myriad possibilities. Begin by
analysis of the F.JOB.TIMES file data. Look for:-
1) Jobs in which the vast majority of the elapsed time is spent in the
SELECT. This may identify where indexes need to be introduced. Look at
the complexity of the SELECT statement. It may need to be simplified
in order for indexes to be used - depending on the DBMS and driver
version.
2) Jobs which have a small select time, a small number of records
selected, and a large total elapsed time. This may identify jobs with
very few but very large transactions.
3) Jobs which take a long time and are not really necessary.
4) Jobs which take longer than you would expect. This may be down to
lock-contention.
5) Jobs which sometimes run quickly and sometimes slowly. Look for
jobs which were running concurrently when the job ran slowly. This may
be due to lock-contention.

This is just a start. Be sure that local developments are written
correctly to fit in with the multi-threaded nature of CoB processing.

[Shiva Shankar]

Hi,

 

* Please look out for any select on STMT.ENTRY or huge files, SELECT on such files is bad programming.

 

only two issue strike me

 

one: ridiculous programming

two: processor capacity

 

people are not aware of the size of the data present in the system.

 

as mike said, this is a huge question!!!

[ADENIYI KAREEM]

Hi ,

 

How did u resolve SELECT on huge file ...like STMT.ENTRY then...?

Lagos,Nigeria
+2348023382844

[Shiva Shankar]

 

if there is an option of coding in a way where the SELECT on the STMT.ENTRY can be averted, that would be a good option.

i do not see any one approving a tech spec where there is a select on the stmt.entry file.

 

[raphael olapade]

hi,

 

This is issue is general when STMT.ENTRY table is huge; however; in the interim LOCATE verb can be use by setting

via the date in the STMT.ENTRY(i.e BOOKING.DATE) and this will be specify in the STANDARD.SELECTION. It is this date that system will use to select the range/particular date you want.

 

Note run your routine at backend and spool it to a file name in the specified folder.

 

 

Thanks

 

Segun 

 

Tel No:234-1-(0)7028030184

[Luis Lascano]

Yudi,

 

Apart from all answers you have received to such generic question, and narrowing down to STMT.ENTRY, best way to get STMT.ENTRY ids is by using API routine EB.ACCT.ENTRY.LIST where you have to specify account number and period of entries you need.

Normally, at end of month COB you will need entries created during last month.

Selecting all entries from STMT.ENTRY during daily or monthly COBs does not have a business sense but if this is the case, a solution could be creating an external index on this table if you need to specify a selection parameter.

 

Below a sample using this API:

 

            ACCT.ID   = 121212

            INI.DATE  = 20080101

            END.DATE  = 20080131

            YID.LIST  = ''

            INI.BAL   = ''

            ERR       = ''

            CALL EB.ACCT.ENTRY.LIST(ACCT.ID,INI.DATE,END.DATE,YID.LIST,INI.BAL,ERR)

 

YID.LIST will have Stmt Entries IDs.

 

Jim,

 

Do we still need to specify [T24] on subject?

 

Regards,

 

Luis

---

[Jim Idle]

 

There is always the BASIC select statement and avoid sorting unless the output demands it. But that is also what indexes are for too and jBC has lots of good indexing commands to help you there.

 

Jim

 

 

From: Shiva Shankar [mailto:aqua.sh...@gmail.com]
Sent: Thursday, February 21, 2008 5:21 AM
To: jB...@googlegroups.com
Subject: Re: what to do to reduce COB time?

 

 

if there is an option of coding in a way where the SELECT on the STMT.ENTRY can be averted, that would be a good option.

[Treg Taylor]

All,

 

Trying to run JBuildSLib and receiving unresolved external symbol _JediSystemLock

 

What lib do I need to include?

 

 

Thanks in advance,

Treg

 

 

 

Win 2000

Jbase 3.4.2.5

[Jim Idle]

 

Do we still need to specify [T24] on subject?

 

Regards,

 

Luis

 

 

Yes please, however there is a lot of overlap. Optimizing a machine and programs is fairly generic for instance, so really [T24] is when you want to know something about using/administering/etc T24 itself and would not be of interest to anyone who does not use T24.

 

Jim

[Jim Idle]

jBuildSLib is essentially a wrapper for the system linker command (usually ld or the cc/gcc front end to that). Hence when you build a library, you need to supply all other libraries that may contain symbols/functions that you are going to call, otherwise the linker cannot determine what libraries load at runtime.

 

In jBASE 3.4.2.5 on Windows, jBuildSLib should automatically provide the jBASE libraries, so it looks like you are trying to call a function that is no longer provided.  This function was part of the jEDI stuff, but I think that for some reason it had a name change, I think to just JediLock, looking at my own drivers. You only need this in your driver if you don’ want jBASE to manage record locks for you.

 

You can use the –v (might be –V) option on this command to see what is going on under the hood, but I think you just need to check the .h files to verify that.

 

Jim

[Treg Taylor]

Thanks Jim,

 

That worked!

 

Treg

 

---

[Shiva Shankar]

 

SELECT on the STMT.ENTRY is only one case, it might not be the problem with your system.

 

To stress on the point again, people are not aware of the size of the data involved. The number of transactions involved, the programming standards involved, the server capacity.

 

It would be a wise option if you get deeper into it and find out where the problem is?

 

To start with as jim said you can check the JOB.PROGRESS and check out which job is taking much time to execute and scan the routine.
 

On 2/21/08, Jim Idle <ji...@temporal-wave.com> wrote:

 

Do we still need to specify [T24] on subject?

 

Regards,

 

Luis

 

 

Yes please, however there is a lot of overlap. Optimizing a machine and programs is fairly generic for instance, so really [T24] is when you want to know something about using/administering/etc T24 itself and would not be of interest to anyone who does not use T24.

 

Jim

[bdf]

HI, shiva thanks for the advice

the job that currently taken the longest time is
BNK/REPORT.PRINT.REPORTING-EB.EOD.REPORT.PRINT
01:40:43 - 02:38:37

any suggestion what to do with this jobs?

thanks

yudi

On Feb 22, 11:11 am, "Shiva Shankar" <aqua.shanka...@gmail.com> wrote:
> SELECT on the STMT.ENTRY is only one case, it might not be the problem with
> your system.
>
> To stress on the point again, people are not aware of the size of the data
> involved. The number of transactions involved, the programming standards
> involved, the server capacity.
>
> It would be a wise option if you get deeper into it and find out where the
> problem is?
>
> To start with as jim said you can check the JOB.PROGRESS and check out which
> job is taking much time to execute and scan the routine.
>

> On 2/21/08, Jim Idle <j...@temporal-wave.com> wrote:
>
>
>
> > *Do we still need to specify [T24] on subject?*
>
> > * *
>
> > *Regards,*
>
> > * *
>
> > *Luis*

[Phuong Do Duc]

I think you can not do anything with this job. Unless, you dont want to have any system reports after daily COB, you can set it run by monthly in BATCH file.

2008/2/22, bdf <yd8...@gmail.com>:

[praveen kumar]

Hi Rifai,
 
This is a single threaded JOB....Please check how many reports you are printing daily? is it really required? Instead of spooling the report at the end you can spool immediatly since other batches are multithreaded it will save some time.
 
ENQ CRF.REPORT.RETAIL IMMEDIATE---- this will spool the report immediately instead at the end.

 

Regards,
Praveen

[Shiva Shankar]

hi praveen,

 

is it mandatory that all these reports should be generated during the cob process??

is it not possible that the reports be generated online???

 

regards
 

[praveen kumar]

Hi Shiva,

CRB and CRF reports can be generated online. Local reports they have to check because some time they may require data as of before start of day.

Regards,
Praveen             

[praveen kumar]

Hi Rafai,

 

            I think we can reduce the COB time, can you mail me the screen shot of the batch BNK/EB.STAT.PRINT.

 

Regards,

Praveen

  

 

On 2/22/08, rifai tes <yd8...@gmail.com> wrote:

HI, praveen

the job that currently taken the longest time is
BNK/REPORT.PRINT.REPORTING-EB.EOD.REPORT.PRINT
01:40:43 - 02:38:37

any suggestion what to do with this jobs?

thanks & regards

yudi

On Thu, Feb 21, 2008 at 11:20 AM, praveen kumar <pav...@gmail.com> wrote:

hi,

 

   can you pls let me know which job is taking more time?

 

regards,

praveen

 

[Santosh]

If any particular job is taking long time. Go to the PGM.FILE of that paricular job and in 3rd field ADDITIONAL.INFO put as .NTX. This may help you out!

--
Mobile: 966-567493144

[Mike Preece]

On Mar 12, 11:33 am, Santosh <santosh.kancha...@gmail.com> wrote:
> If any particular job is taking long time. Go to the PGM.FILE of that
> paricular job and in 3rd field ADDITIONAL.INFO put as .NTX. This may help
> you out!
>
>

LOL! That's a pretty bold statement! User beware... do you really want
to turn off transaction management for that job? Will the boss thank
you if something bad happens and he finds that partial transaction
updates have been committed?

[Mike Preece]

On Feb 21, 4:20 pm, "Jim Idle" <j...@temporal-wave.com> wrote:
> There is always the BASIC select statement and avoid sorting unless the
> output demands it. But that is also what indexes are for too and jBC has
> lots of good indexing commands to help you there.
>
> Jim
>

Just discovered this in rereading the thread. Be advised that changing
from a jQL SELECT to a BASIC SELECT will make a big difference if your
data resides within jBASE, but not if you're using Oracle or some
other RDBMS. jBASE (and "Pick-like" systems generally) work on lists.
RDBMSs work on result-sets. Fundamentally very different. If you want
to begin processing a list all you really need is the start point. If
you're dealing with a result-set you need to wait until you have a
complete set before you can begin.

[Jim Idle]

> -----Original Message-----
> From: jB...@googlegroups.com [mailto:jB...@googlegroups.com] On Behalf

> Of Mike Preece
> Sent: Wednesday, March 12, 2008 5:08 AM
> To: jBASE
> Subject: Re: what to do to reduce COB time?
>
>
>
>

If this statement was correct, then there would be a lot of very angry web users waiting on complete result sets from Oracle/SQL Server etc ;-).

It depends on how you ask for the result set and what you ask of the result set. For instance if you ask for the difference between the average (non-moving) value of a field and the record value, then you may need to traverse the result set to calculate such a thing, but even then, the server can perform a local query to find the aggregate it needs to produce the average on the fly. Simple WHERE clauses do not need to produce the entire result set before they can give back result sets, which are generally paged if you know there will be a lot of results.

Don't forget that the indexes will (or at least did) work on a jEDI driver, hence if your select is based on index chopping, you will get a list and the submission to Oracle will be an individual read. Depending on what you are doing this can be faster. So, it depends on what the Oracle drivers try to do with a jQL query vs producing a list and iterating. At one point, we were trying to get the Oracle drivers to intercept queries and use local indexes if available, which is just what I am advocating here.

Other than someone in corporate headquarters has mandated Oracle with as much qualification to do so as perform deep sea diving, I see no reason that most users should use such a database really. It just adds more complications. There are other reasons to want to use such things though ;-), such as all your staff are already trained in Oracle.

Jim

[Phuong Do Duc]

Hi Praveen.

 

Where can I run ENQ CRF.REPORT.RETAIL IMMEDIATE?
 

Regards
 

2008/3/10, praveen kumar <pav...@gmail.com>:

[praveen kumar]

Hi,

Below shown screen shot is self-explanatory.

               BATCH ENTRY SEE
 
     BATCH PROCESS..... BKB/RE.PRINT.DEAL.STAFF
 ------------------------------------------------------------------------------
   1 BATCH.STAGE....... R800                REPORTING
   3 PROCESS.STATUS.... 0                   READY
   4 BATCH.ENVIRONMENT. F                   FOREGROUND
   6. 1 JOB.NAME....... RE.OUTPUT.EXTRACT
   8. 1 FREQUENCY...... D                   DAILY
  11. 1. 1 DATA........ STAFF
  12. 1 JOB.STATUS..... 0                   READY
  13. 1 LAST.RUN.DATE.. 03 MAR 2008
   6. 2 JOB.NAME....... EB.PRINT
   7. 2. 1 VERIFICATION RE.OUTPUT.EXTRACT
   8. 2 FREQUENCY...... D                   DAILY
  11. 2. 1 DATA........ ENQ CRB.REPORT.STAFF IMMEDIATE

Thanks & Regards

Praveen

 

[Phuong Do Duc]

Hi Praveen,

 

I really dont understand, in this example of BATCH record named BKB/RE.PRINT.DEAL.STAFF, in 11. 2. 1 DATA........ ENQ CRB.REPORT.STAFF IMMEDIATE.

 

1 - How relation between BKB/RE.PRINT.DEAL.STAFF and CRB.REPORT.STAFF.

2 - Does this setup apply only for Single Thread?

3 - If I have batch record like below:

 

         BATCH PROCESS..... BNK/REPORT.PRINT.REPORTING
 ------------------------------------------------------------------------------
   1 BATCH.STAGE....... R999                REPORTING
   2 DEFAULT.PRINTER...

   3 PROCESS.STATUS.... 0                   READY
   4 BATCH.ENVIRONMENT. F                   FOREGROUND

   5 DEPARTMENT.CODE...
   6. 1 JOB.NAME....... EB.EOD.REPORT.PRINT
   7. 1. 1 VERIFICATION

   8. 1 FREQUENCY...... D                   DAILY

   9. 1 NEXT.RUN.DATE..
  10. 1 PRINTER.NAME...
  11. 1. 1 DATA........

  12. 1 JOB.STATUS..... 0                   READY

  13. 1 LAST.RUN.DATE.. 13 MAR 2008
  14. 1 JOB.MESSAGE....
  15. 1 USER...........
  16 RECORD.STATUS.....
 

Which value can I set in 11.1.1 field?

 

Thanks and Regards.

 

2008/3/13, praveen kumar <pav...@gmail.com>:

2008/3/10, praveen kumar <pav...@gmail.com>:

Home phone: 84 4 5543349

[praveen kumar]

Hi Phuong,

                I will explain you the concept of using IMMEDIATE in the batch records.

Option 1 – When IMMEDAITE is not mentioned in the batch records.

In any batch record if you are printing any reports using either ENQ/ROUTINE/RPG then EB.PRINT job will process and will write to EB.REPORT. Once the reporting reaches the Final stage, the last batch in the reporting-BNK/REPORT.PRINT.REPORTING will start printing all the reports reading from EB.REPORT using the routine EB.EOD.REPORT.PRINT

How do we test this?

This time I will take the core batch record- modify the batch records BNK/EU.PL.REALISED.REPORT & BNK/REPORT.PRINT.REPORTING & create TSA services so that you can run them as services.

Clear everything from &HOLD&.

BATCH ENTRY SEE

     BATCH PROCESS..... BNK/EU.PL.REALISED.REPORT

 ------------------------------------------------------------------------------

   1 BATCH.STAGE....... R100                REPORTING

   3 PROCESS.STATUS.... 0                   READY

   4 BATCH.ENVIRONMENT. F                   FOREGROUND

   6. 1 JOB.NAME....... EB.PRINT

   8. 1 FREQUENCY...... D                   DAILY

  11. 1. 1 DATA........ ENQ EU.FX.PL.TODAY

  12. 1 JOB.STATUS..... 0                   READY

  13. 1 LAST.RUN.DATE.. 03 NOV 2000

  17 CURR.NO........... 1

  18. 1 INPUTTER....... SY_COMP.IN.BATCH

  19. 1 DATE.TIME...... 07 MAY 02 18:57

  20 AUTHORISER........ 12_S.DONEY

  21 CO.CODE........... GB-001-0001         NEWMASTER 5.005

  22 DEPT.CODE......... 1

Run the above shown batch record as a service and check LIST FBNK.EB.REPORT, you can see a record with the enquiry name EU.FX.PL.TODAY.

Now check LIST &HOLD& nothing will be there Now start the tSA agent of BNK/REPORT.PRINT.REPORTING, it will read FBNK.EB.REPORT and will write to &HOLD&.

Now check LIST &HOLD& you will find one record.

Options 2 – When IMMEDAITE is mentioned in the batch records

In any batch record if you are printing any reports using either ENQ/ROUTINE/RPG then EB.PRINT job will process and will write to EB.REPORT. If ENQ/RTN/RPG contains 'IMMEDIATE' along with its name then it will be printed immediately.

How do we test this?

This time modify the batch records BNK/EU.PL.REALISED.REPORT & create TSA services so that you can run them as services.(This time do not create tSA service for BNK/REPORT.PRINT.REPORTING)

Clear everything from &HOLD&.

BATCH ENTRY SEE

 

     BATCH PROCESS..... BNK/EU.PL.REALISED.REPORT

 ------------------------------------------------------------------------------

   1 BATCH.STAGE....... R100                REPORTING

   3 PROCESS.STATUS.... 0                   READY

   4 BATCH.ENVIRONMENT. F                   FOREGROUND

   6. 1 JOB.NAME....... EB.PRINT

   8. 1 FREQUENCY...... D                   DAILY

  11. 1. 1 DATA........ ENQ EU.FX.PL.TODAY IMMEDIATE

  12. 1 JOB.STATUS..... 0                   READY

  13. 1 LAST.RUN.DATE.. 03 NOV 2000

  17 CURR.NO........... 1

  18. 1 INPUTTER....... SY_COMP.IN.BATCH

  19. 1 DATE.TIME...... 07 MAY 02 18:57

  20 AUTHORISER........ 12_S.DONEY

  21 CO.CODE........... GB-001-0001         NEWMASTER 5.005

  22 DEPT.CODE......... 1

 

Run the above shown batch record as a service and check &HOLD&. You will find one record.

Please check the PGM.FILE of EB.EOD.REPORT.PRINT to know whether BNK/REPORT.PRINT.REPORTING is a single threaded or multithreaded.

We use IMMEDIATE usually to print the report at that point of stage & if BNK/REPORT.PRINT.REPORTING is a single threaded in your Globus release then YES it does help to reduce the COB time. If BNK/REPORT.PRINT.REPORTING is taking more time in spite it's multithreaded then please check RE.STAT.REQUEST records. If you have any record with more reports attached to it, then it will be better to split it.

Regards,

Praveen

[Phuong Do Duc]

Hi Praveen,

 

Your guide is clear, I will try with our Test server. then if I have any problems, I will contact you again.

 

Thank you for sharing me your exprience.

 

Regards.

 

2008/3/14, praveen kumar <pav...@gmail.com>:

Hi Phuong,

                I will explain you the concept of using IMMEDIATE in the batch records.

Option 1 - When IMMEDAITE is not mentioned in the batch records.

In any batch record if you are printing any reports using either ENQ/ROUTINE/RPG then EB.PRINT job will process and will write to EB.REPORT. Once the reporting reaches the Final stage, the last batch in the reporting-BNK/REPORT.PRINT.REPORTING will start printing all the reports reading from EB.REPORT using the routine EB.EOD.REPORT.PRINT

How do we test this?

This time I will take the core batch record- modify the batch records BNK/EU.PL.REALISED.REPORT & BNK/REPORT.PRINT.REPORTING & create TSA services so that you can run them as services.

Clear everything from &HOLD&.

BATCH ENTRY SEE

     BATCH PROCESS..... BNK/EU.PL.REALISED.REPORT

 ------------------------------------------------------------------------------

   1 BATCH.STAGE....... R100                REPORTING

   3 PROCESS.STATUS.... 0                   READY

   4 BATCH.ENVIRONMENT. F                   FOREGROUND

   6. 1 JOB.NAME....... EB.PRINT

   8. 1 FREQUENCY...... D                   DAILY

  11. 1. 1 DATA........ ENQ EU.FX.PL.TODAY

  12. 1 JOB.STATUS..... 0                   READY

  13. 1 LAST.RUN.DATE.. 03 NOV 2000

  17 CURR.NO........... 1

  18. 1 INPUTTER....... SY_COMP.IN.BATCH

  19. 1 DATE.TIME...... 07 MAY 02 18:57

  20 AUTHORISER........ 12_S.DONEY

  21 CO.CODE........... GB-001-0001         NEWMASTER 5.005

  22 DEPT.CODE......... 1

Run the above shown batch record as a service and check LIST FBNK.EB.REPORT, you can see a record with the enquiry name EU.FX.PL.TODAY.

Now check LIST &HOLD& nothing will be there Now start the tSA agent of BNK/REPORT.PRINT.REPORTING, it will read FBNK.EB.REPORT and will write to &HOLD&.

Now check LIST &HOLD& you will find one record.

Options 2 - When IMMEDAITE is mentioned in the batch records

[mike ryder]

Cor blimey - more advice than you know what to do with - and nobody
has even bothered to ask "which T24 release" because that is *rather*
relevant, nor how many application servers, nor what database.

The job in question is reporting - in which case NTX is irrelevent
because txn control is turned off for reporting jobs (Hmm - that seems
to depend on whether you mean R4, R5 or R6)

You need to examine the tables that tell you how long things are
taking - JOB.TIMES and PRINT.TIMES (i think that is the name??) and
determine which print jobs are consuming the time. Are they T24 core
jobs or something of your own invention? I have seen localdev
reporting jobs that take 3 hours because it is not doing *just*
reporting - it is processing loads of transactions and then reporting
on what it has processed.

I have given some ideas here to indicate that this isn't a trivial
question that can be solved in a newsgorup (How can I make Vista work?
- change registry setting xyzfix) because the problem is complex and
needs analysis before proposing a solution (as in my example - the
wrong solution because of inadequate analysis - the correct solution
is to delete Vista and install Ubuntu)

To the OP - you need to request support from your temenos account
manager for a tech consultant.

Ooops - end of tirade

Mike

[Truong Giang Huynh]

Hi,

 

You should search all your database so that you know which file should be resized, then you resize all of them.

 

Please tell me about SAN, your model SAN is using, also vendor.

 

If you're runing on AIX, you can use Nmon tool for monitoring I/O so that how many percent I/O it is using on the COB time.

 

 

Regards,

HTGIANG

T24 Admin at Sacombank in Viet Nam