Document

Download Report

Transcript Document 

Windows NT/2000
Performance and Capacity
Key Metrics
Jerry L. Rosenberg
SRM Associates, Ltd.
e-mail: [email protected]
http://www.srmassoc.com
2000 SRM Associates, Ltd.
Capacity and Performance

Performance Tuning


Identify bottlenecks and take immediate corrective
action
Capacity Planning

Anticipate future bottlenecks and avoid them


Understand current usage
Project for expected changes
2000 SRM Associates, Ltd.
NT Data

There is a very rich set of data available
in NT from Performance Monitor

It is critical to avoid data overload

Particularly in time critical performance situations,
it is vital to pinpoint the problem and correct it
rapidly
2000 SRM Associates, Ltd.
Still only categories of data
Processor
 I/O
 Memory


What about Network ?
2000 SRM Associates, Ltd.
Performance Standard
If the processor is not excessively busy
and work is completing on schedule, all
is well.
 If work is not meeting service levels and
the processor use is excessive, begin
by examining the processor.

2000 SRM Associates, Ltd.
Performance Standard

If work is not meeting service levels and
the processor use is not excessive, look
to I/O or memory as the problem.

Beware: It is not quite that simple.
2000 SRM Associates, Ltd.
Processor

Processor: % Processor Time

< 50%
80% = Danger
System: % Total Processor Time
 Processor: Interrupts/sec



< 3500 for Pentium
System: Processor Queue Length

<2
2000 SRM Associates, Ltd.
2000 SRM Associates, Ltd.
2000 SRM Associates, Ltd.
2000 SRM Associates, Ltd.
2000 SRM Associates, Ltd.
2000 SRM Associates, Ltd.
8.97
8.08
7.18
6.28
5.38
4.49
3.59
2.69
1.79
0.90
0.00
0
1
MN15.ttotlcpu_average
2000 SRM Associates, Ltd.
2
3
4
5
6
7
17
18
MN13.ttotlcpu_average
19
20
21
22
23
Processor – Other Metrics
System:Context Switches/sec
 Process:% Processor Time, % Priv.
Time, % User Time, Priority Base
 Thread:% Processor Time, Priority
Current

2000 SRM Associates, Ltd.
Memory

Soft Faults


Page/faults/sec
Hard Faults

Pages/sec
2000 SRM Associates, Ltd.
6/21/2000
6/21/2000
6/22/2000
6/22/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
6/21/2000
47771.40
42994.26
38217.12
33439.98
28662.84
23885.70
19108.56
14331.42
9554.28
4777.14
0.00
pagflts_sum
2000 SRM Associates, Ltd.
memp_sum
92.77
83.49
74.21
64.94
55.66
46.38
37.11
27.83
18.55
9.28
0.00
0
1
2
pagoutsec_average
2000 SRM Associates, Ltd.
3
4
5
6
7
8
paginsec_average
9
10
11
12
13
14
15
pagrdsec_average
16
17
18
19
20
21
22
pagwrtsec_average
23
2000 SRM Associates, Ltd.
Paging Standard
Memory: Pages/sec * PhysicalDisk:
Average Disk sec/Transfer = amount of
disk time spent on paging activity during
the interval.
 Should not exceed 10%
 20% = thrashing

2000 SRM Associates, Ltd.
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
10/7/2000
0.000048
0.000046
0.000044
Percent of Time Spent on Paging
0.000042
0.000040
0.000038
0.000036
0.000034
0.000032
0.000030
0.000028
0.000026
0.000024
0.000022
0.000020
0.000018
0.000016
0.000014
0.000012
0.000010
0.000008
0.000006
0.000004
0.000002
0.000000
Pagechk
2000 SRM Associates, Ltd.
Memory – Other Metrics

Memory:Available Bytes

Process:Page Faults/sec, Working Set
2000 SRM Associates, Ltd.
I/O

Physical


Logical


Actual Device
Partition
Need diskperf –y or diskperf -ye
2000 SRM Associates, Ltd.
I/O

% Disk Time


Current Disk Queue Length


< 55%
<2
Avg. Disk Bytes/Transfer

Big numbers are good
2000 SRM Associates, Ltd.
I/O – Other Metrics
Memory:Pages/sec
 Physical Disk:Average Disk Bytes
Read, Disk Bytes/sec
 Processor:% Processor Time,
Interrupts/sec

2000 SRM Associates, Ltd.
33601.21
30241.09
26880.97
23520.84
20160.72
16800.60
13440.48
10080.36
6720.24
3360.12
0.00
0
1
2
3
ldbpw_average
2000 SRM Associates, Ltd.
4
5
6
7
8
9
10
11
12
13
14
ldbpr_average
15
16
17
18
19
20
21
22
23
Network

Complex and configuration dependent




Protocols
NICs
Network Applications
Topology
Maybe multiple protocol stacks
 Hint: monitor Redirector object

2000 SRM Associates, Ltd.
NIC example

NIC:

Output Queue Length


Bytes Total/sec


<2
Should not be high if OQL < 2
If collisions greater than 10%, Network issue
2000 SRM Associates, Ltd.
2000 SRM Associates, Ltd.
Network – Other Metrics






Network Segment:% Net Util., Total Bytes
Received/sec
Memory: Pages/sec
Logical Disk:% Free Space
Paging File:% Peak Usage
Physical Disk: % Disk Time, Avg. Q Len.
Processor:% Processor Time
2000 SRM Associates, Ltd.
452057.53
406851.78
361646.03
316440.27
271234.52
226028.77
180823.01
135617.26
90411.51
45205.75
0.00
0
1
srvbr_average
2000 SRM Associates, Ltd.
2
3
4
5
6
7
17
srvbt_average
18
19
20
21
22
23
Performance
Concentrate on the area of failure
 Limit the metrics analyzed to key
counters
 Have historical data

2000 SRM Associates, Ltd.
Capacity
Build an historical data base
 Review the report periodically for trend
changes
 Model possible effects

2000 SRM Associates, Ltd.
Jerry Rosenberg
Strategic Resource Management
Associates, Ltd.
www.srmassoc.com
516 433-1817
[email protected]
2000 SRM Associates, Ltd.