oracle awr
DESCRIPTION
Oracle AWR. Prepared For: XYZ CorpData Collected on 2011-09-07. Agenda. This Storage Assessment is organized in the following sections: Introduction Summary IO and Load Details IO and Load Details by Instance (if RAC) Top 10 Tablespaces Top 10 Wait Events – IO and Non-IO Related - PowerPoint PPT PresentationTRANSCRIPT
1© Copyright 2010 EMC Corporation. All rights reserved.
Oracle AWRPrepared For: XYZ Corp
Data Collected on 2011-09-07
2© Copyright 2010 EMC Corporation. All rights reserved.
Agenda
• This Storage Assessment is organized in the following sections:
– Introduction– Summary– IO and Load Details– IO and Load Details by Instance (if RAC)– Top 10 Tablespaces– Top 10 Wait Events – IO and Non-IO Related– Burst Analysis
• The following additional files were provided:– Details.xlsx: Additional information to provide further
detail
3© Copyright 2010 EMC Corporation. All rights reserved.
Introduction - Methodology
• The disk estimates in this report are calculated by analyzing the RAID-adjusted IOPS (read + write) and estimating the number of required spindles to sustain the workload. These sizing estimates use EMC's ratings for sustained IOPS, not burst, for each drive speed.
• These estimates are performance based estimates based on the sample data supplied. These estimates do not consider database capacity, capacity growth or specific EMC array capabilities. These estimates make up one part of the recommended EMC architecture.
• This analysis does not include additional spindles required for standard best-practice database components, such as redo, archive logs, backups to disk, replication and clones.
4© Copyright 2010 EMC Corporation. All rights reserved.
Introduction - Drive Calculations
• For RAID-5 and RAID-10 adjusted IOPs– RAID-5 Adjusted IOPS=Physical read IO requests per second +
4*(Physical write IO requests per second + Redo writes per second)– RAID-10 Adjusted IOPS=Physical read IO requests per second +
2*(Physical write IO requests per second + Redo writes per second)
• Note: the metrics used for RAID adjusted IOPs exclude archive log, RMAN backup or restore, and Flashback database IOPs. RAID-5 or RAID-6 10k RPM or 7200 RPM RAID groups or pools are usually sufficient for the sequential IOPs generated by these processes
• For the RAID-5 and RAID-10 drive counts, the relevant adjusted RAID IOPs figure is divided by the IOPs rating for each drive type: 180 IOPs for 15k RPM and 2500 IOPs for EFDs.
5© Copyright 2010 EMC Corporation. All rights reserved.
Introduction - Additional Calculations• The following calculations are used throughout this report
– % Physical Read IO to Total IO = 100 * Physical read IO requests per second / (Physical read IO requests per second + Physical write IO requests per second + Redo writes per second)
– % Percentage physical reads to logical (total) reads = 100*(physical reads per sec / logical reads per sec). These are Oracle data block reads, not read IO calls.
– Read Bandwidth (MB/sec) = Physical reads total bytes per second / (1024*1024)
– Write Bandwidth (MB/sec) = Physical writes total bytes per second / (1024*1024)
• Note: This report shows both 95th percentiles and peak values. When less than 20 samples (AWR reports) are provided, 95th percentile will be mathematically equivalent to the peak value.
6© Copyright 2010 EMC Corporation. All rights reserved.
Input Summary
Database Name
Instance Host Software Version
RAC Start Date End Date # Files Analyzed
Default Block Size
(KB)
xyz xyz1 host01 Oracle 10.2.0.4.0
YES 2011-09-06 00:00
2011-09-07 00:00
96 8
xyz xyz2 host02 Oracle 10.2.0.4.0
YES 2011-09-06 00:00
2011-09-07 00:00
96 8
xyz xyz3 host03 Oracle 10.2.0.4.0
YES 2011-09-06 00:00
2011-09-07 00:00
96 8
7© Copyright 2010 EMC Corporation. All rights reserved.
IOPS and Drive Estimates
Metric IOPS Required EFD Devices Required 15K Spindles
RAID-10 Adjusted IOPS - Peak 11,165 5 63
RAID-5 Adjusted IOPS - Peak 12,717 6 71
RAID-10 Adjusted IOPS - 95th 9,569 4 54
RAID-5 Adjusted IOPS - 95th 11,918 5 67
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
2,0004,0006,0008,000
10,00012,00014,000
RAID-5 IOPS RAID-10 IOPS
IOP
S
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
20
40
60
80RAID-5 15k # Drives RAID-10 15k # Drives
# D
rives
8© Copyright 2010 EMC Corporation. All rights reserved.
IOPS and Drive Estimates By Instance
Name RAID-10 IOPS - max RAID-5 IOPS - max RAID-10 15k # Drives - max RAID-5 15k # Drives - max
xyz:xyz1 5,270 5,978 29.28 33.21
xyz:xyz2 3,605 5,582 20.03 31.01
xyz:xyz3 3,397 3,814 18.87 21.19
These maximums may occur at different times and should not be added to get a total drive estimate.
9© Copyright 2010 EMC Corporation. All rights reserved.
RAID Adjusted IOPS
9/6/2011 9/6/20110
1,000
2,000
3,000
4,000
5,000
6,000
xyz:xyz1 xyz:xyz2 xyz:xyz3
RA
ID-1
0 I
OP
S
9/6/2011 9/6/20110
1,000
2,000
3,000
4,000
5,000
6,000
7,000
xyz:xyz1 xyz:xyz2 xyz:xyz3
RA
ID-5
IO
PS
10© Copyright 2010 EMC Corporation. All rights reserved.
System Summary
Metric Value Comments
# Databases 1 database 3 instances
Database Name xyz Instances: xyz1, xyz2, xyz3
Database Version 10.2.0.4.0
RAID-10 Estimated # 15K Drives 63 (54) Peak (95th)
RAID-5 Estimated # 15K Drives 71 (67) Peak (95th)
RAID-10 Estimated # EFD Drives 5 (4) Peak (95th)
RAID-5 Estimated # EFD Drives 6 (5) Peak (95th)
RAID-10 Adjusted IOPS 11165 (9569) Peak (95th)
RAID-5 Adjusted IOPS 12717 (11918) Peak (95th)
Physical Reads/Total Physical IO 89% (100%) Average (Peak)
Physical Reads/Logical Reads 10% (37%) Average (Peak)
Read Bandwidth (MB/sec) 346 Peak Value
Write Bandwidth (MB/sec) 51 Peak Value
11© Copyright 2010 EMC Corporation. All rights reserved.
IO and Load Details
12© Copyright 2010 EMC Corporation. All rights reserved.
Database Workload(IOPS)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
2,000
4,000
6,000
8,000
10,000
12,000
Phys Read IOs/sec Phys Read IOs/sec - 95th
Phys R
ead I
Os/s
ec
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
200
400
600
800
1,000
1,200
Phys Write IOs/sec Phys Write IOs/sec - 95th
Phys W
rite
IO
s/s
ec
13© Copyright 2010 EMC Corporation. All rights reserved.
Database Workload(Redo)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
100
200
300
400
500
600
700
Redo W
rite
s/s
ec
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
2,000,000
4,000,000
6,000,000
8,000,000
10,000,000
12,000,000
14,000,000
Redo B
yte
s/s
ec
14© Copyright 2010 EMC Corporation. All rights reserved.
Physical Read Percentages
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
102030405060708090
100
Physic
al R
ead I
O/T
ota
l IO
%
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
102030405060708090
100
Physic
al R
ead/L
ogic
al
Read %
15© Copyright 2010 EMC Corporation. All rights reserved.
Database Bandwidth
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
50
100
150
200
250
300
350
400
Physical Read Total MB/sec Physical Write Total MB/sec
MB
/sec
16© Copyright 2010 EMC Corporation. All rights reserved.
Transactions, Executes and User Calls
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
1,0002,0003,0004,0005,0006,000
Tra
nsacti
ons/s
ec
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
2,0004,0006,0008,000
10,00012,00014,000
Execute
s/s
ec
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
5,000
10,000
15,000
20,000
25,000
User
Calls/s
ec
17© Copyright 2010 EMC Corporation. All rights reserved.
IO and Load Details by Instance
18© Copyright 2010 EMC Corporation. All rights reserved.
Instance Workload(IOPS) (xyz:xyz1)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
1,000
2,000
3,000
4,000
5,000
6,000
Phys Read IOs/sec Phys Read IOs/sec - 95th
Phys R
ead I
Os/s
ec
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
50100150200250300350400450
Phys Write IOs/sec Phys Write IOs/sec - 95th
Phys W
rite
IO
s/s
ec
19© Copyright 2010 EMC Corporation. All rights reserved.
Instance Workload(Redo) (xyz:xyz1)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
50
100
150
200
250
Redo W
rite
s/s
ec
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
1,000,0002,000,0003,000,0004,000,0005,000,0006,000,0007,000,0008,000,0009,000,000
10,000,000
Redo S
ize/s
ec
20© Copyright 2010 EMC Corporation. All rights reserved.
Instance Workload(IO Ratios) (xyz:xyz1)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
102030405060708090
100
Physic
al R
ead I
O/T
ota
l IO
%
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
102030405060708090
100
Physic
al R
ead/L
ogic
al
Read %
21© Copyright 2010 EMC Corporation. All rights reserved.
Instance Workload(IOPS) (xyz:xyz2)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
500
1,000
1,500
2,000
2,500
3,000
Phys Read IOs/sec Phys Read IOs/sec - 95th
Phys R
ead I
Os/s
ec
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
100
200
300
400
500
600
700
Phys Write IOs/sec Phys Write IOs/sec - 95th
Phys W
rite
IO
s/s
ec
22© Copyright 2010 EMC Corporation. All rights reserved.
Instance Workload(Redo) (xyz:xyz2)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
50100150200250300350400450
Redo W
rite
s/s
ec
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
1,000,0002,000,0003,000,0004,000,0005,000,0006,000,0007,000,0008,000,0009,000,000
10,000,000
Redo S
ize/s
ec
23© Copyright 2010 EMC Corporation. All rights reserved.
Instance Workload(IO Ratios) (xyz:xyz2)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
102030405060708090
100
Physic
al R
ead I
O/T
ota
l IO
%
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
102030405060708090
100
Physic
al R
ead/L
ogic
al
Read %
24© Copyright 2010 EMC Corporation. All rights reserved.
Instance Workload(IOPS) (xyz:xyz3)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
500
1,000
1,500
2,000
2,500
3,000
3,500
Phys Read IOs/sec Phys Read IOs/sec - 95th
Phys R
ead I
Os/s
ec
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
50
100
150
200
250
300
Phys Write IOs/sec Phys Write IOs/sec - 95th
Phys W
rite
IO
s/s
ec
25© Copyright 2010 EMC Corporation. All rights reserved.
Instance Workload(Redo) (xyz:xyz3)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
10
20
30
40
50
60
70
Redo W
rite
s/s
ec
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
500,000
1,000,000
1,500,000
2,000,000
2,500,000
3,000,000
Redo S
ize/s
ec
26© Copyright 2010 EMC Corporation. All rights reserved.
Instance Workload(IO Ratios) (xyz:xyz3)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
102030405060708090
100
Physic
al R
ead I
O/T
ota
l IO
%
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
102030405060708090
100
Physic
al R
ead/L
ogic
al
Read %
27© Copyright 2010 EMC Corporation. All rights reserved.
Instance Bandwidth (xyz1)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
20
40
60
80
100
120
140
Physical Read Total MB/sec Physical Write Total MB/sec
MB
/sec
28© Copyright 2010 EMC Corporation. All rights reserved.
Instance Bandwidth (xyz2)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
50
100
150
200
250
300
350
Physical Read Total MB/sec Physical Write Total MB/sec
MB
/sec
29© Copyright 2010 EMC Corporation. All rights reserved.
Instance Bandwidth (xyz3)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
10
20
30
40
50
60
70
80
90
Physical Read Total MB/sec Physical Write Total MB/sec
MB
/sec
30© Copyright 2010 EMC Corporation. All rights reserved.
Transactions, Executes and User Calls by Instance (xyz1)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
500
1,000
1,500
2,000
Tra
nsacti
ons/s
ec
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
5001,0001,5002,0002,5003,0003,500
Execute
s/s
ec
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
2,000
4,000
6,000
8,000
10,000
User
Calls/s
ec
31© Copyright 2010 EMC Corporation. All rights reserved.
Transactions, Executes and User Calls by Instance (xyz2)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
5001,0001,5002,0002,5003,000
Tra
nsacti
ons/s
ec
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
2,000
4,000
6,000
8,000
Execute
s/s
ec
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
2,0004,0006,0008,000
10,00012,00014,000
User
Calls/s
ec
32© Copyright 2010 EMC Corporation. All rights reserved.
Transactions, Executes and User Calls by Instance (xyz3)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
200
400
600
800
Tra
nsacti
ons/s
ec
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
2,0004,0006,0008,000
10,00012,000
Execute
s/s
ec
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:360
500
1,000
1,500
2,000
User
Calls/s
ec
33© Copyright 2010 EMC Corporation. All rights reserved.
Oracle Host CPU Utilization (xyz:xyz1)
34© Copyright 2010 EMC Corporation. All rights reserved.
Oracle Host CPU Utilization (xyz:xyz2)
35© Copyright 2010 EMC Corporation. All rights reserved.
Oracle Host CPU Utilization (xyz:xyz3)
36© Copyright 2010 EMC Corporation. All rights reserved.
Tablespaces
37© Copyright 2010 EMC Corporation. All rights reserved.
Tablespaces
• Oracle tablespaces are logical containers holding various database segments, such as of tables, indices or their partitions, undo or rollback, system meta-data, and temporary sort area.
• Underlying a tablespace are one or more files or devices
• IO contention is indicated where tablespaces are showing average read times in milliseconds (Av Rd(ms)) equal to or greater than 20 milliseconds, though with modern array caching, 10 milliseconds or more is becoming the indicator of IO contention.
• EFDs show their best performance gains with small-block random-read OLTP-like workloads. See Blocks Per Read for read size, where a block is an Oracle data block. The default Oracle data block size is on the Input Summary slide.
• Databases showing IO contention via high values for Av Rd ms and showing small block random reads (8-16KB) via Blocks per Read, in conjunction with the database block size, are ideal candidates for EFDs.
38© Copyright 2010 EMC Corporation. All rights reserved.
Tablespace Metrics at IO Peak (xyz1)
Start Date Tablespace # Reads Read IOPS Avg Rd (ms) Blocks per Read # Writes Write IOPS
2011-09-06 10:30 XYZ_DOC570_LOB 4,238,561 4685 0 1 3,959 4
2011-09-06 10:30 XYZ_DOC450_DATA 220,228 243 8 1 1 0
2011-09-06 10:30 XYZ_DOC543_LOB 43,664 48 0 1 11,392 13
2011-09-06 10:30 XYZ_DOC543_DATA 28,618 32 6 13 12,076 13
2011-09-06 10:30 XYZ_DOC570_DATA 11,118 12 11 1 763 1
2011-09-06 10:30 XYZ_DOC570_INDEX 6,684 7 11 1 1,347 1
2011-09-06 10:30 TEMP_DATA 4,029 4 0 1 3,916 4
2011-09-06 10:30 XYZ_DOC631_DATA 5,180 6 0 1 2 0
2011-09-06 10:30 XYZ_DOC543_RS0001_LOB 3,367 4 0 1 3 0
2011-09-06 10:30 XYZ_DOC593_DATA 1,935 2 10 1 17 0
39© Copyright 2010 EMC Corporation. All rights reserved.
Tablespace Metrics at IO Peak (xyz2)
Start Date Tablespace # Reads Read IOPS Avg Rd (ms) Blocks per Read # Writes Write IOPS
2011-09-06 10:30 XYZ_DOC560_LOB 930,361 1028 0 1 168 0
2011-09-06 10:30 TEMP_DATA 282,875 313 0 1 61,229 68
2011-09-06 10:30 XYZ_DOC676_LOB 428 0 0 1 258,583 286
2011-09-06 10:30 XYZ_DOC675_LOB 214,226 237 0 1 493 1
2011-09-06 10:30 XYZ_DOC669_INDEX 74,424 82 8 1 104,086 115
2011-09-06 10:30 XYZ_DOC675_DATA 91,328 101 5 1 2,391 3
2011-09-06 10:30 XYZ_DOC675_INDEX 52,540 58 9 1 19,560 22
2011-09-06 10:30 XYZ_DOC631_LOB 56,325 62 2 1 0 0
2011-09-06 10:30 XYZ_DOC560_INDEX 32,986 36 6 1 16,814 19
2011-09-06 10:30 XYZ_DOC543_RS0001_DATA 37,767 42 7 1 1 0
40© Copyright 2010 EMC Corporation. All rights reserved.
Tablespace Metrics at IO Peak (xyz3)
Start Date Tablespace # Reads Read IOPS Avg Rd (ms) Blocks per Read
# Writes Write IOPS
2011-09-06 10:30 XYZ_DOC508_DATA 1,521,133 1681 2 2 2 0
2011-09-06 10:30 XYZ_DOC631_DATA 164,303 182 1 5 0 0
2011-09-06 10:30 XYZ_DOC661_DATA 89,173 99 3 1 76 0
2011-09-06 10:30 XYZ_DOC570_LOB 40,605 45 0 1 5,462 6
2011-09-06 10:30 XYZ_DOC543_RS0001_DATA 38,016 42 6 1 61 0
2011-09-06 10:30 XYZ_DOC675_INDEX 18,486 20 6 1 2,393 3
2011-09-06 10:30 XYZ_DOC661_LOB 15,466 17 0 1 3,049 3
2011-09-06 10:30 XYZ_DOC543_LOB 11,779 13 1 1 3,169 4
2011-09-06 10:30 XYZ_DOC661_DOC661_DATA 14,231 16 5 1 531 1
2011-09-06 10:30 XYZ_DOC675_DATA 12,411 14 5 1 1,208 1
41© Copyright 2010 EMC Corporation. All rights reserved.
Top 10 Tablespaces (by Total IOs) (xyz1)
9/6/2011 9/6/2011 9/7/20110
1,000
2,000
3,000
4,000
5,000XYZ_DOC570_LOB XYZ_DOC543_DATA XYZ_DOC629_INDEXTEMP_DATA XYZ_DOC629_DATA XYZ_DOC450_DATAXYZ_DOC543_LOB XYZ_DOC543_RS0001_DATA XYZ_DOC516_DATAXYZ_DOC614_DATA
Read +
Wri
te
IOP
S
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:36 0:000
5
10
15
20
Avg R
d (
ms)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:36 0:000
5
10
15
20
Blo
cks p
er
Read
Due to the changing locality of database activity, the listed tablespaces can show discontinuities.
42© Copyright 2010 EMC Corporation. All rights reserved.
Top 10 Tablespaces (by Total IOs) (xyz2)
9/6/2011 9/6/2011 9/7/20110
500
1,000
1,500
2,000
2,500XYZ_DOC636_DATA XYZ_DOC644_DATA TEMP_DATA XYZ_DOC669_INDEXXYZ_DOC490_DATA XYZ_DOC630_DATA XYZ_DOC644_LOB XYZ_DOC516_DATAXYZ_DOC644_INDEX XYZ_DOC508_DATA
Read +
Wri
te
IOP
S
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:36 0:000
5
10
15
20
Avg R
d (
ms)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:36 0:000
5
10
15
20
Blo
cks p
er
Read
43© Copyright 2010 EMC Corporation. All rights reserved.
Top 10 Tablespaces (by Total IOs) (xyz3)
9/6/2011 9/6/2011 9/7/20110
500
1,000
1,500
2,000
2,500XYZ_DOC490_DATA XYZ_DOC508_DATA TEMP_DATA XYZ_DOC521_DATAXYZ_DOC570_LOB XYZ_DOC543_DATA XYZ_DOC644_LOB XYZ_DOC629_DATAXYZ_DOC614_DATA XYZ_DOC629_INDEX
Read +
Wri
te
IOP
S
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:36 0:000
5
10
15
Avg R
d (
ms)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:36 0:0005
101520253035
Blo
cks p
er
Read
44© Copyright 2010 EMC Corporation. All rights reserved.
Top Wait Events
45© Copyright 2010 EMC Corporation. All rights reserved.
Wait Events
• The wait events analyzed here are from the Top 5 Timed/Waited Events section of the AWR reports.
• Response time can be thought of as the sum of service time, i.e. CPU time, plus wait time.
• Gaps indicate that an event was not in the Top 5 for a particular sample.
• Oracle uses the terms sequential and scattered to mean the opposite. "Sequential" implies random IOs, and "scattered" implies sequential IOs.
• Use these slides to evaluate indicators of IO contention, as well as suitability for EFDs.
• EFDs show their best performance gains with random read IOs, indicated as "sequential reads". EFDs show good to moderate gains with sequential read IOs, indicated as "scattered reads".
• The column '% Total Call Time' represents the percentage of time during processing an Oracle process was waiting on this IO related wait event and 'Av Wait MS' is the average process wait time in milliseconds.
46© Copyright 2010 EMC Corporation. All rights reserved.
Wait Events(IO) (xyz1)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:36 0:000
5
10
15
20
25
30
System I/O:control file sequential read User I/O:db file sequential readUser I/O:direct path read User I/O:db file scattered readCommit:log file sync System I/O:log file parallel writeUser I/O:db file parallel read User I/O:direct path read tempSystem I/O:log file sequential read User I/O:direct path write
Avg W
ait
(m
s)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:36 0:000
102030405060708090
100
% T
ota
l C
all T
ime
Extracted from the "Top 5 Timed Events" section of AWR/StatsPack report.
47© Copyright 2010 EMC Corporation. All rights reserved.
Wait Events(Non-IO) (xyz1)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:36 0:000
100
200
300
400
500
600
Other:enq: WF - contention Cluster:gc cr disk readCluster:gc cr grant 2-way Cluster:gc cr multi block requestOther:reliable message Cluster:gc current block 3-wayOther:PX Deq Credit: send blkd Network:SQL*Net more data to clientOther:enq: TX - contention Applicatio:enq: TX - row lock contentionApplicatio:enq: TM - contention Concurrenc:enq: TX - index contention
Avg W
ait
(m
s)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:36 0:000
102030405060708090
100
% T
ota
l C
all T
ime
48© Copyright 2010 EMC Corporation. All rights reserved.
Wait Events(IO) (xyz2)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:36 0:000
10
20
30
40
50
60
70
80
System I/O:control file sequential read System I/O:db file parallel writeUser I/O:db file sequential read System I/O:RMAN backup & recovery I/OSystem I/O:log file parallel write User I/O:direct path readUser I/O:Data file init write User I/O:db file scattered readUser I/O:direct path write User I/O:direct path read tempUser I/O:direct path write temp Commit:log file sync
Avg W
ait
(m
s)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:36 0:000
50
100
150
200
250
300
350
% T
ota
l C
all T
ime
49© Copyright 2010 EMC Corporation. All rights reserved.
Wait Events(Non-IO) (xyz2)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:36 0:000
100
200
300
400
500
600
Cluster:gc cr disk read Other:change tracking file synchronoOther:DFS lock handle Other:enq: CF - contentionOther:PX Deq Credit: send blkd Cluster:gc cr multi block requestOther:enq: WF - contention Other:reliable messageApplicatio:enq: TX - row lock contention Cluster:gc buffer busyOther:enq: TS - contention
Avg W
ait
(m
s)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:36 0:000
102030405060708090
100
% T
ota
l C
all T
ime
50© Copyright 2010 EMC Corporation. All rights reserved.
Wait Events(IO) (xyz3)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:36 0:000
5
10
15
20
25
System I/O:control file sequential read User I/O:db file sequential readUser I/O:direct path read User I/O:db file scattered readUser I/O:direct path read temp User I/O:direct path write tempUser I/O:read by other session System I/O:log file parallel writeSystem I/O:log file sequential read User I/O:db file parallel read
Avg W
ait
(m
s)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:36 0:000
102030405060708090
100
% T
ota
l C
all T
ime
51© Copyright 2010 EMC Corporation. All rights reserved.
Wait Events(Non-IO) (xyz3)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:36 0:000
500
1,000
1,500
2,000
2,500
Cluster:gc cr disk read Other:enq: WF - contention Other:change tracking file synchronoOther:reliable message Cluster:gc current block 2-way Cluster:gc cr grant 2-wayOther:PX Deq Credit: send blkd Other:PX Deq: Table Q Get Keys Cluster:gc cr multi block requestCluster:gc current block busy Cluster:gc cr block 2-way Concurrenc:cursor: pin S wait on XConcurrenc:library cache lock Other:gcs log flush sync Other:cr request retry
Avg W
ait
(m
s)
0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:36 0:000
102030405060708090
100
% T
ota
l C
all T
ime
52© Copyright 2010 EMC Corporation. All rights reserved.
Burst Analysis
53© Copyright 2010 EMC Corporation. All rights reserved.
Peak BurstPeak IO event was 2011-09-06 10:30 to 10:50. +-6 samples is
09:00 thru 12:20
54© Copyright 2010 EMC Corporation. All rights reserved.
Burst by Instance
9/6/2011 9:00:00 9/6/2011 10:12:00 9/6/2011 11:24:000
1,000
2,000
3,000
4,000
5,000
6,000
xyz1 xyz2 xyz3
Phys I
Os/s
ec
55© Copyright 2010 EMC Corporation. All rights reserved.
Top 5 Tablespaces during Peak Burst
9/6/2011 9:00:00 9/6/2011 10:12:00 9/6/2011 11:24:000
1,000
2,000
3,000
4,000
5,000XYZ1:PM_DOC450_DATA XYZ1:PM_DOC570_LOB XYZ2:PM_DOC636_DATAXYZ3:PM_DOC508_DATA XYZ2:PM_DOC630_DATA
Read +
Wri
te
IOP
S
9:00 9:28 9:57 10:26 10:55 11:24 11:520.00
2.00
4.00
6.00
8.00
10.00
Avg R
d (
ms)
9:00 9:28 9:57 10:26 10:55 11:24 11:5205
101520253035
Blo
cks p
er
Read
56© Copyright 2010 EMC Corporation. All rights reserved.
Wait Events during Burst(IO) (xyz1)
9:00 9:28 9:57 10:26 10:55 11:24 11:520123456789
10
System I/O:control file sequential read User I/O:db file scattered readUser I/O:db file sequential read User I/O:direct path readSystem I/O:log file parallel write User I/O:direct path read tempCommit:log file sync
Avg W
ait
(m
s)
9:00 9:28 9:57 10:26 10:55 11:24 11:520
102030405060708090
100
System I/O:control file sequential read User I/O:db file scattered readUser I/O:db file sequential read User I/O:direct path readSystem I/O:log file parallel write User I/O:direct path read tempCommit:log file sync
% T
ota
l C
all T
ime
57© Copyright 2010 EMC Corporation. All rights reserved.
Wait Events during Burst(Non-IO) (xyz1)
9:00 9:28 9:57 10:26 10:55 11:240
100
200
300
400
500
600
Cluster:gc cr disk read Other:PX Deq Credit: send blkdNetwork:SQL*Net more data to client Other:enq: TX - contentionApplicatio:enq: TX - row lock contention Applicatio:enq: TM - contention
Avg W
ait
(m
s)
9:00 9:28 9:57 10:26 10:55 11:240
102030405060708090
100
Cluster:gc cr disk read Other:PX Deq Credit: send blkdNetwork:SQL*Net more data to client Other:enq: TX - contentionApplicatio:enq: TX - row lock contention Applicatio:enq: TM - contention
% T
ota
l C
all T
ime
58© Copyright 2010 EMC Corporation. All rights reserved.
Wait Events during Burst(IO) (xyz2)
9:00 9:28 9:57 10:26 10:55 11:24 11:520
1
2
3
4
5
6
7
8
System I/O:control file sequential read User I/O:db file sequential readUser I/O:direct path write User I/O:direct path readCommit:log file sync System I/O:log file parallel writeUser I/O:db file scattered read User I/O:direct path read temp
Avg W
ait
(m
s)
9:00 9:28 9:57 10:26 10:55 11:24 11:520
102030405060708090
100
System I/O:control file sequential read User I/O:db file sequential readUser I/O:direct path write User I/O:direct path readCommit:log file sync System I/O:log file parallel writeUser I/O:db file scattered read User I/O:direct path read temp
% T
ota
l C
all T
ime
59© Copyright 2010 EMC Corporation. All rights reserved.
Wait Events during Burst(Non-IO) (xyz2)
9:00 9:28 9:57 10:26 10:55 11:24 11:520
100
200
300
400
500
600
Other:reliable message Applicatio:enq: TX - row lock contentionCluster:gc buffer busy
Avg W
ait
(m
s)
9:00 9:28 9:57 10:26 10:55 11:24 11:520
102030405060708090
100
Other:reliable message Applicatio:enq: TX - row lock contentionCluster:gc buffer busy
% T
ota
l C
all T
ime
60© Copyright 2010 EMC Corporation. All rights reserved.
Wait Events during Burst(IO) (xyz3)
9:00 9:28 9:57 10:26 10:55 11:24 11:520
2
4
6
8
10
12
14
User I/O:db file sequential read User I/O:direct path read User I/O:direct path read tempUser I/O:read by other session User I/O:db file scattered read System I/O:log file parallel write
Avg W
ait
(m
s)
9:00 9:28 9:57 10:26 10:55 11:24 11:520
102030405060708090
100
User I/O:db file sequential read User I/O:direct path read User I/O:direct path read tempUser I/O:read by other session User I/O:db file scattered read System I/O:log file parallel write
% T
ota
l C
all T
ime
61© Copyright 2010 EMC Corporation. All rights reserved.
Wait Events during Burst(Non-IO) (xyz3)
9:00 9:28 9:57 10:26 10:55 11:24 11:520
50100150200250300350400450
Other:PX Deq Credit: send blkd Cluster:gc current block busyOther:enq: WF - contention Cluster:gc cr grant 2-way
Avg W
ait
(m
s)
9:00 9:28 9:57 10:26 10:55 11:24 11:520
102030405060708090
100
Other:PX Deq Credit: send blkd Cluster:gc current block busyOther:enq: WF - contention Cluster:gc cr grant 2-way
% T
ota
l C
all T
ime