slide 1
DB2 V10 upgrade experience @ Colruyt
slide 2
DB2 V10 upgrade experience @ Colruyt
• Who are we • Migration plan • Issues • Performance • New features
slide 3
Who are we...
France
Belgium
Other Wholesale & Food service Retail
Luxembourg
Colex
slide 4
Who are we... • DB2 Architecture
DBP1
DBS2
DBT2
DBA1
DBI1
DBP2
DBS1
DBT1
DBA2
DBE1
Legend: • Production
• Development
• Test
• Training
• Acceptence
• DBA Sandbox
Galileï (z196-2817) 6 CPU & 2 zIIP 6251 mips 96 GB
Newton (z196-2817) 5 CPU & 2 zIIP
5300 mips 96 GB
DB2P
DB2S
DB2T
DB2A
DB2I
DB2E
XPC01
XOC01
XAC01
slide 5
Migration plan • New customer since 2008 started in DB2 V9
– No existing migration plans
• Setup DBA sandbox (DB2I) environment (January 2012)
• Pre-migration tasks (January 2012 – February 2012) – Upgrade Third party Tools – Check for necessary V9 PTF’s – Check for incompatibilities / ZPARM changes – DSNTIJPA
• Migration to Conversion Mode (February 2012 – June 2012)
– With preventive maintenance prior to migration DB2P
• Migration to New-Function Mode (August 2012 – October 2012)
slide 6
Migration plan Premigration Tasks
– Migration Third Party Product
• IBM Utility Suite – Log Analysis Tool – Query Monitor Tool – Administration Tool – Tivoli Performance Monitor
• DB2 Connect – Upgraded from V9.1 to V10.1
» V9.7 FP3a required for new functions
• Replication Tools • Windows Client Tools
– Datastudio – AQT / Toad
slide 7
Migration plan Premigration Tasks
– Check for necessary V9 PTF’s
• Current on maintenance (December 2011)) – Check for incompatibilities / ZPARM changes
• A few ZPARM changes were done before migration started – Altered minimal/default values
• Modified DSN3SATH • Deprecated DSNAEXP move to new EXPLAIN privileges (TODO 2013)
– DSNTIJPA • Some PLANs that contained DBRMs • Migration of EXPLAIN tables
– SMS managed Catalog & Directory • Already implemented in V9 through SMS rules • Extended addressability
slide 8
Migration plan SMS Environment Setup
DB2 Catalog
DB2 Directory
Active Log Copy 1
DB2 BSDS 1
Active Log Copy 2
DB2 BSDS 2
User Data
Tool Data
Workfile Database
Archive Logs
GDB2PCP1 GDB2PCP2
CDB2PCP1
DDB2EXT
CDB2PCP2
GDB2PCP3
CDB2PCP3
GDB2PCP4
CDB2PCP4
GDB2PCP5
CDB2PCP5
GDB2PCP7
CDB2PCP7
Backups
Storagegroup
Storageclass
Dataclass
DDB2EXT
slide 9
Migration plan Migration to Conversion Mode
– DSNTIJXZ
• Update default ZPARM input member (DSNTIDxx) • Buffer Pool Settings
– Ensure that no DB2 work is running • Cancelled utilities !!
– DSNTIJMV / DSNTIJRW • New / Modified WLM environments
– DSNTIJEX • Recompile security exit DSN3SATH
– DSNTIJTC • Migrate to Conversion Mode
– Rebind all packages • Rebind with PLANMGMT (EXTENDED)
– Available since DB2 V9 4 hours
slide 10
Migration plan Issues in Conversion Mode
– Issues with IBM Tools
• IBM Administration Tool generating faulty scripts • IBM Query Monitor displaying incorrect elapsed times
– Utility jobs abending with 00C4 • Every time dynamic sort space allocation occurred • Workaround: specify SORTDEVT & SORTNUM • Problem CA-ACF2 interface
– IDLE THREAD TIMEOUT • ABEND=04E REASON=00D3003B • Threads were kept active even after commit
– Are forced out of DB2 (IDHTHOIN = 120 sec) – UK78127 was applied
• Fault in planning – REBIND job ran when member 1 was migrated – WLM choose to run on member 2 – Rebind all fixed the issue
• Migration plan was adapted
slide 11
Migration plan Issues in Conversion Mode
– High CPU consumption in MSTR STC - SRB
• Attributed to internal memory management for statistics • PTF UK80191 resolves the problem
– Currently being implemented – High CPU consumption in MSTR STC – TCB
• Introduced after applying z/OS maintenance • DB2 gathers additional real storage information • z/OS fix UA63988 should reduce the CPU time required to gather this information
– Currently being implemented – Drop of + 70% TCB CPU time in DB2A
slide 12
Migration plan High CPU consumption in MSTR STC – TCB TIME
MSTR - Total Consumption
0
50
100
150
200
250
300
30/0
1/12
6/02
/12
13/0
2/12
20/0
2/12
27/0
2/12
5/03
/12
12/0
3/12
19/0
3/12
26/0
3/12
2/04
/12
9/04
/12
16/0
4/12
23/0
4/12
30/0
4/12
7/05
/12
14/0
5/12
21/0
5/12
28/0
5/12
4/06
/12
11/0
6/12
18/0
6/12
25/0
6/12
2/07
/12
9/07
/12
16/0
7/12
23/0
7/12
30/0
7/12
6/08
/12
13/0
8/12
20/0
8/12
27/0
8/12
3/09
/12
10/0
9/12
17/0
9/12
24/0
9/12
1/10
/12
8/10
/12
15/1
0/12
22/1
0/12
29/1
0/12
5/11
/12
12/1
1/12
19/1
1/12
26/1
1/12
3/12
/12
WEEK
CPU
seco
nds
SSAS_TCB_TIME SSAS_SRB_TIME SSAS_PRE_CP_SRB SSAS_PRE_IIP_SRB TOTAL_CPU
slide 13
Migration plan Migration to New-Function Mode
– DSNTIJEN
• Migrate to Enabling New-Function Mode • Converts the catalog tables
– DSNTIJNF • Migrate to New-Function Mode
– DSNTIJNG • Update DSNHDECP load module
– DSNTIJRT/DSNTIJRV • Install and verify DB2 supplied routines
– DSNTIJCL / DSNTIJMS • ODBC & JDBC support
– Rebind all packages • Rebind with PLANMGMT (EXTENDED)
– REORG Catalog & Directory 2 hours
slide 14
Migration plan Issues in New-Function Mode
– DB2 connect Unlimited Edition V10.1
• SQL1598N An attempt to connect to the database server failed because of a licensing problem. • The db2connectactivate utility activates the license and grants access to SYSIBM.DB2CON_LIC package
– SYSIBM.DB2CON_LIC package: » Will update a local file conlic.bin (if last entry is not older then 7 days)
• CA-ACF2 interface as external security package – Grant to SYSIBM.DB2CON_LIC package was not done by db2connectactivate utility
• After granting EXECUTE on package everything OK
slide 15
Performance • SQL Performance comparison:
– Weekly sum CPU consumption by thread type
– Daily Average CPU consumption per SQL statement by
thread type
– Thread types: • DRDA • CICS • BATCH (PL/I)
slide 16
DRDA performance
0
10.000
20.000
30.000
40.000
50.000
60.000
14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48
WEEK
CPU
seco
nds
0
2
4
6
8
10
12
14
Millio
nsSQ
L STM
T (x 1
.000.0
00)
CLASS2_CPU_AGENT CLASS2_IIP_CPU OCCURENCES
slide 17
DRDA performance (weekdays only)
0,000000
0,000100
0,000200
0,000300
0,000400
0,000500
0,000600
0,000700
2/04/1
2
9/04/1
2
16/04
/12
23/04
/12
30/04
/12
7/05/1
2
14/05
/12
21/05
/12
28/05
/12
4/06/1
2
11/06
/12
18/06
/12
25/06
/12
2/07/1
2
9/07/1
2
16/07
/12
23/07
/12
30/07
/12
6/08/1
2
13/08
/12
20/08
/12
27/08
/12
3/09/1
2
10/09
/12
17/09
/12
24/09
/12
1/10/1
2
8/10/1
2
15/10
/12
22/10
/12
29/10
/12
5/11/1
2
12/11
/12
19/11
/12
26/11
/12
AVG CPU/SQL 30 per. Mov. Avg. (AVG CPU/SQL)
slide 18
DRDA Performance (weekdays only)
• DRDA Performance conclusion: V9
NFM CM NFMAVG_CPU_SQL 0,000174 0,000159 0,000172
% 100,00% 91,13% 98,88%
AVG_IIP_SQL 0,000273 0,000229 0,000253% 100,00% 83,87% 92,81%
V10
93,44%
86,53%
slide 19
CICS performance
0
500
1.000
1.500
2.000
2.500
3.000
3.500
4.000
4.500
14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48
WEEK
CPU
seco
nds
0
100.000
200.000
300.000
400.000
500.000
600.000
700.000
800.000
900.000
1.000.000
SQL S
TMT (
x 1.00
0.000
)
CLASS2_CPU_AGENT OCCURENCES
slide 20
CICS performance (weekdays only)
0,000000
0,000200
0,000400
0,000600
0,000800
0,001000
0,001200
2/04/1
2
9/04/1
2
16/04
/12
23/04
/12
30/04
/12
7/05/1
2
14/05
/12
21/05
/12
28/05
/12
4/06/1
2
11/06
/12
18/06
/12
25/06
/12
2/07/1
2
9/07/1
2
16/07
/12
23/07
/12
30/07
/12
6/08/1
2
13/08
/12
20/08
/12
27/08
/12
3/09/1
2
10/09
/12
17/09
/12
24/09
/12
1/10/1
2
8/10/1
2
15/10
/12
22/10
/12
29/10
/12
5/11/1
2
12/11
/12
19/11
/12
26/11
/12
3/12/1
2
AVG CPU/STMT 30 per. Mov. Avg. (AVG CPU/STMT)
slide 21
CICS Performance (weekdays only)
• CICS Performance conclusion: V9
NFM CM NFMAVG_CPU_SQL 0,000766 0,000721 0,000774
% 100,00% 94,13% 101,04%
V10
96,12%
slide 22
BATCH performance
0
10.000
20.000
30.000
40.000
50.000
60.000
70.000
80.000
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48
WEEK
CPU
seco
nds
0
500
1.000
1.500
2.000
2.500
3.000
Millio
nsSQ
L STM
T (x 1
.000.0
00)
CLASS7_CPU_AGENT CLASS7_IIP_CPU SQL_STMTS_ISSUED
slide 23
BATCH performance
0,000000
0,000005
0,000010
0,000015
0,000020
0,000025
0,000030
0,000035
0,000040
0,000045
0,000050
2/03/12
12/03/12
22/03/121/04/12
11/04/12
21/04/121/05/12
11/05/12
21/05/12
31/05/12
10/06/12
20/06/12
30/06/12
10/07/12
20/07/12
30/07/129/08/12
19/08/12
29/08/128/09/12
18/09/12
28/09/128/10/12
18/10/12
28/10/127/11/12
17/11/12
27/11/120
500.000.000
1.000.000.000
1.500.000.000
2.000.000.000
2.500.000.000
3.000.000.000
3.500.000.000
4.000.000.000
4.500.000.000
5.000.000.000
AVG CPU/STMT SQL_STMTS_ISSUED 30 per. Mov. Avg. (AVG CPU/STMT)
slide 24
BATCH Performance
• BATCH Performance conclusion:
V9NFM CM NFM
AVG_CPU_SQL 0,000022 0,000180 0,000012% 100,00% 83,54% 57,36%
V10
75,80%
slide 25
Subsystem Performance - MSTR
• MSTR CPU consumption increased: – After CM:
• Due to DB2 internal memory managemnt process – PREEMPT SRB – Fixed in UK80191 currently being applied
• +45% for TOTAL_MSTR_CPU
– After Z/OS maintenance: (25/08/2012) • z/OS PTF UA60825
– Fixed in UA63988 applied in XOC01
• +300% in TCB_MSTR_CPU
slide 26
Subsystem Performance - MSTR
MSTR - Total Consumption DB2P
0
500
1000
1500
2000
2500
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48
WEEK
CP
U s
econ
ds
SSAS_TCB_TIME SSAS_SRB_TIME SSAS_PRE_CP_SRB SSAS_PRE_IIP_SRB
slide 27
Subsystem Performance – DBM1
• DIST CPU consumption moved to zIIP: – DBM1_SRB_CPU
• -50%
– DBM1_SRB_IIP • +200%
– DBM1_TOTAL_CPU
• Netto saving of 40%
slide 28
Subsystem Performance – DBM1
DBM1 - Total Consumption DB2P
0
5000
10000
15000
20000
25000
30000
35000
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48
WEEK
CPU
seco
nds
DSAS_TCB_TIME DSAS_SRB_TIME DSAS_PRE_CP_SRB DSAS_PRE_IIP_SRB
slide 29
Subsystem Performance – DIST
DDF - Total Consumption DB2P
0
10000
20000
30000
40000
50000
60000
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48
WEEK
CPU
seco
nds
DDF_TCB_TIME DDF_SRB_TIME DDF_PRE_CP_SRB DDF_PRE_IIP_SRB
slide 30
RUNSTATS CPU/IIP TIME DB2 RUNSTATS
0
200
400
600
800
1.000
1.200
1.400
1.600
14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48
WEEK
CP
U S
EC
0
2000
4000
6000
8000
10000
12000
OC
CU
RR
EN
CE
S
SUM_CL1_CPU SUM_CL1_IIP OCCURRENCES
slide 31
New-features
• REBIND options APREUSE/APCOMPARE • New security privileges • Temporal Data • 1MB FRAME size • Index INCLUDE • …
slide 32
Questions ?