Job Management Partner 1/Performance Management - Remote Monitor for Platform Description, User's Guide and Reference

Appendix A.2 Disk space requirements

The required disk space depends on the number of records to be used to store performance data.

The following subsections provide estimates of the disk space required for PFM - RM for Platform.

Organization of this subsection

(1) Disk space requirement for the entire system

(2) Disk space requirement for the Store database

(1) Disk space requirement for the entire system

The disk space requirement for PFM - RM for Platform depends on the file sizes for the following data:

Remote Monitor Store database

Number of PFM - RM for Platform instances

Number of hosts monitored by PFM - RM for Platform

Agent logs

The following table shows how to estimate the disk space requirement for the entire system.

Table A-3 Disk space requirement for the entire system

Status of PFM - RM for Platform Disk space requirement (megabytes)

Windows Server 2003 / Windows Server 2008 / Linux

During installation^# 5

During operation a + b

Legend:

a:

Disk space required during installation

b:

Total disk space required for each instance.

The formula for estimating the disk space required for a single instance follows:

100 + c + d + f

c:

Disk space required for instance logs:

Default: 24 megabytes (3 megabytes x 8 files)

Maximum: 256 megabytes (32 megabytes x 8 files)

d:

Total disk space required for each monitored host.

The formula for estimating the disk space required for a single monitored host follows:

1 + e

e:

Disk space required for logs per monitored host:

Default: 12 megabytes (3 megabytes x 4 files)

Maximum: 128 megabytes (32 megabytes x 4 files)

f:

Total disk space required for the Store database.

This is the total of the disk space required for each monitored host and the disk space required for group agents. For details about the disk space requirement for the Store database, see (2) Disk space requirement for the Store database.

#

During installation, you need twice as much disk space as the size of the program.

Status of PFM - RM for Platform	Disk space requirement (megabytes)
Windows Server 2003 / Windows Server 2008 / Linux
During installation^#	5
During operation	a + b

(2) Disk space requirement for the Store database

This subsection presents the formulas for estimating the disk space required for the Store database of PFM - RM for Platform and provides an example of making such an estimate.

(a) Formulas

This subsection describes how to estimate the disk space requirement, number of files, number of directories, and number of files that are opened by the Store service.

n Disk space requirement

The disk space requirement for the Store database is the sum of the disk space requirement for each record type. For the PI record type, the disk space requirement is the sum of the disk space requirement for each summary type.

The following shows the formula for estimating the disk space requirement for each record type:
Disk space requirement for each record type (bytes)
X = {(e + 2) x f + (d + 60) x {((e + 2) x f)/(65,250 - d) + 1}^#} x a/b x (c + 1) x 1.1
Note

a:

The value depends on the record type and summary type. See Table A-4.

b:

The value depends on the record type and summary type. See Table A-4.

c:

Value set as the retention period for historical data. For details about the default retention period, see Table A-5. The units depend on the record type and summary type. For details about the units, see Table A-4.

d:

Size of the fixed part of each record that collects historical data. For details about the size of the fixed part of each record, see the record size in 5. Records.

e:

Size of the variable part of each record that collects historical data. For details about the size of the variable part of each record, see the record size in 5. Records.

f:

Number of instances for each record that collects historical data (1 for a single-instance record). For details about how to calculate the number of record instances, see Table A-6. If a calculation method is provided for a record, use that method.

If the number of instances is 2 or greater, use the next even multiple of 4. For example, if the number of instances is 2, use 4 as the value of f. If the number of instances is 13, use 16 as the value of f. If the number of instances is 1, use 1 as the value of f.

For the number of instances for a group agent, use the highest number of monitored instances that have been registered. For example, if the number of instances of PI_CPU for monitoring target A is 5 and the number of instances of PI_CPU for monitoring target B is 2, use 5 as the number of instances for the group agent.

#

Round off the result of {((e + 2) x f)/(65,250 - d) + 1}.

The following table shows the values to be set in variables a, b, and c.

Table A-4 Values to be set in variables a, b, and c

Record type Summary type Variable

a b c

PD -- 1,440 g/60 Retention period (days)

PI Minute 1,440 1 + (g - 1)/60^# Retention period (days)

Hour 24 1 + (g - 1)/3,600^# Retention period (days)

Day 7 1 + (g - 1)/86,400^# Retention period (weeks)

Week 1 1 + (g - 1)/604,800^# Retention period (weeks)

Month 1 1 + (g - 1)/2,592,000^# Retention period (months)

Year 1 1 + (g - 1)/31,622,400^# Retention period (years)

Legend:

--: Not applicable

g: Value set as the historical data collection interval (seconds)

#

Round off the calculation result of variable b for the PI record type.

The following table lists the retention periods for historical data.

Table A-5 Retention periods for historical data

Record type Data type Retention period

PD -- 7 days

PI Minute-by-minute 1 day

Hourly 7 days

Daily 54 weeks

Weekly 54 weeks

Monthly 1 year

Yearly Unlimited

Legend:

--: Not applicable

The following table shows how to estimate the number of record instances.

Table A-6 How to estimate the number of record instances

Record How to estimate the number of instances Checking required

PD 1 (because this is a single-instance record) --

PI 1 (because this is a single-instance record) --

PI_CPU Number of processors on the monitored host + 1 (_Total). Y

PI_LDSK Number of logical disks on the monitored host + 1 (_Total). Y

PI_NET Number of network interfaces on the monitored host + 1 (_Total). Y

PI_PDSK Number of physical disks allocated to the monitored host + 1 (_Total). Y

Legend:

--: Not applicable

Y: The number of instances must be checked.

The following describes how to check the number of record instances.

In Windows

Check the number of record instances in the Performance window, which is displayed by choosing Administrative Tools from the Start menu and then choosing Performance. The following table shows the correspondence between the information displayed in the Performance window and the records.

Table A-7 Correspondence between the information displayed in the Performance window and the records

Record Description

PI_CPU Check the number displayed for the instances of the following performance object (including _Total):

Processor

PI_LDSK Check the number displayed for the instances of the following performance object (including _Total):

Logical disk

PI_NET Check the number displayed for the instances of the following performance object + 1 (including _Total):

Network interface

PI_PDSK Check the number displayed for the instances of the following performance object (including _Total):

Physical disk

In UNIX

Execute a command to check the number of record instances. The following table shows the commands used to check the number of record instances.

Table A-8 Commands used to check the number of record instances

Record Commands Description

HP-UX Solaris AIX Linux

PI_CPU sar -Muw 1 1 mpstat -p 1 2 sar -u -P ALL 1 1 mpstat -P The value returned by the command is the number of CPUs + 1.

PI_LDSK df -lk df -lk df -k df -lkP The value returned by the command is the number of file systems + 1.

PI_NET netstat -in netstat -in netstat -in netstat -in The value returned by the command is the number of interfaces + 1.

PI_PDSK iostat iostat -x 1 1 sar -d 1 1 iostat -xkd 1 1 The value returned by the command is the number of disk devices + 1.

n Number of files

The following shows the formula for estimating the number of files (N) that are created for the Store database.
Formula for estimating the number of files (N) created for the Store database:
N = 20 + 2 x (
        (A11 + A12 +...+ A1m + m) +
        (A21 + A22 +...+ A2m + m) +
        (A31 + A32 +...+ A3m + m) +
        (A41 + A42 +...+ A4m + m) +
        (A51 + A52 +...+ A5m + m) +
        (11 x m) +
        (B1 + B2 +...+ Bn + n)
        )
Note

m:

Number of PI records collected

n:

Number of PD records collected

From A11 to A1m:

Value set as the retention period for each minute-by-minute record of the PI record type (days)

From A21 to A2m:

Value set as the retention period for each hourly record of the PI record type (days)

From A31 to 3m:

Value set as the retention period for each daily record of the PI record type (weeks)

From A41 to 4m:

Value set as the retention period for each weekly record of the PI record type (weeks)

From A51 to 5m:

Value set as the retention period for each monthly record of the PI record type (months)

From B1 to Bn:

Value set as the retention period for each record of the PD record type (days)

n Number of directories

The following shows the formula for estimating the number of directories (N) that are created for the Store database.
Formula for estimating the number of directories (N) created for the Store database:
N = 25 + 2 x ((A1max) + (A2max) + (A3max) + (A4max) + (A5max) + 11 + (Bmax))
Note

m:

Number of PI records collected

n:

Number of PD records collected

A1max:

Maximum value set as the retention period for data whose summary type is Minute for a record of the PI record type (days)

A2max:

Maximum value set as the retention period for data whose summary type is Hour for a record of the PI record type (days)

A3max:

Maximum value set as the retention period for data whose summary type is Day for a record of the PI record type (weeks)

A4max:

Maximum value set as the retention period for data whose summary type is Week for a record of the PI record type (weeks)

A5max:

Maximum value set as the retention period for data whose summary type is Month for a record of the PI record type (months)

Bmax:

Maximum value set as the retention period for each record of the PD record type (days)

n Number of files opened by the Store service

The following shows the formula for estimating the number of files that are opened by the Store service.
Formula for estimating the number of files opened by the Store service:
N = 20 + 2 x (6 x m + n)
Note

m:

Number of PI records collected

n:

Number of PD records collected

(b) Example

This subsection describes estimation of the disk space requirement for the Store database for PFM - RM for Platform by way of an example.

n Disk space requirement

This example collects PI_CPU and PD records.

The following describes how to estimate the disk space requirement for the PI_CPU record. The values of variables d through g in the formula for estimating the disk space requirement are shown below. For the meanings of the variables d through g, see (a) Formulas.
d = 1,034 (bytes)
e = 544 (bytes)
f = 4
g = 300 (seconds)
Next, the values of the variables a through c, such as the minute-by-minute and hourly records, and the formulas are described below. For the meanings of the variables a through c, see (a) Formulas.
Minute-by-minute record
a = 1,440
b = 1 + (300 - 1)/60 = 5.98 (round off the value)
c = 3 (days)
The formula is as follows:
X(minute-by-minute) = {(544 + 2) x 4 + (1,034 + 60) x {((544 + 2) x 4)/(65,250 - 1,034) + 1}} x 1,440/5 x (3 + 1) x 1.1
= {2,184 + 1,094 x 1} x 1,267.2
= 3,278 x 1,267.2
= 4,153,881 (bytes) = approximately 4 (megabytes)
Hourly record
a = 24
b = 1 + (300 - 1)/3,600 = 1.08 (round off the value)
c = 3 (days)
The formula is as follows:
X(hourly) = {(544 + 2) x 4 + (1,034 + 60) x {((544 + 2) x 4)/(65,250 - 1,034) + 1}} x 24/1 x (3 + 1) x 1.1
= {2,184 + 1,094 x 1} x 105.6
= 3,278 x 105.6
= 346,156 (bytes) = approximately 0.4 (megabytes)
Daily record
a = 7
b = 1 + (300 - 1)/86,400 = 1.00 (round off the value)
c = 1 (week)
The formula is as follows:
X (daily) = {(544 + 2) x 4 + (1,034 + 60) x {((544 + 2) x 4)/(65,250 - 1,034) + 1}} x 7/1 x (1 + 1) x 1.1
= {2,184 + 1,094 x 1} x 15.4
= 3,278 x 15.4
= 50,481 (bytes) = approximately 0.05 (megabytes)
Weekly record
a = 1
b = 1 + (300 - 1)/604,800 = 1.00 (round off the value)
c = 1 (week)
The formula is as follows:
X(weekly) = {(544 + 2) x 4 + (1,034 + 60) x {((544 + 2) x 4)/(65,250 - 1,034) + 1}} x 1/1 x (1 + 1) x 1.1
= {2,184 + 1,094 x 1} x 2.2
= 3,278 x 2.2
= 7,211 (bytes) = approximately 0.01 (megabytes)
Monthly record
a = 1
b = 1 + (300 - 1)/2,592,000 = 1.00 (round off the value)
c = 1 (month)
The formula is as follows:
X(monthly) = {(544 + 2) x 4 + (1,034 + 60) x {((544 + 2) x 4)/(65,250 - 1,034) + 1}} x 1/1 x (1 + 1) x 1.1
= {2,184 + 1,094 x 1} x 2.2
= 3,278 x 2.2
= 7,211 (bytes) = approximately 0.01 (megabytes)
Yearly record
a = 1
b = 1 + (300 - 1)/31,622,400 = 1.00 (round off the value)
c = 10 (years)
The formula is as follows:
X(yearly) = {(544 + 2) x 4 + (1,034 + 60) x {((544 + 2) x 4)/(65,250 - 1,034) + 1}} x 1/1 x (10 + 1) x 1.1
= {2,184 + 1,094 x 1} x 12.1
= 3,278 x 12.1
= 39,663 (bytes) = approximately 0.04 (megabytes)
From the above, the estimate for PI_CPU is as follows:
X(total) = X(minute-by-minute) + X(hourly) + X(days) + X(weekly) + X(monthly) + X(yearly)
= 4.51 (megabytes)
= approximately 5 (megabytes)
The following describes how to estimate the disk space requirement for the PD record. The values of the variables d through g in the formula for estimating the disk space requirement are shown below. For the meanings of the variables d through g, see (a) Formulas.
a = 1,440
b = 300/60 = 5
c = 7 (days)
d = 2,050 (bytes)
e = 0 (bytes)
f = 1
g = 300 (seconds)
The formula is as follows:
X = {(0 + 2) x 1 + (2,050 + 60) x {((0 + 2) x 1)/(65,250 - 2,050) + 1}} x 1,440/5 x (7 + 1) x 1.1
= {2,112 x 1} x 2,534.4
= 2,112 x 2,534.4
= 5,352,653 (bytes) = approximately 6 (megabytes)
Therefore, the disk space requirement is PI_CPU + PD = 11 (megabytes).

n Number of files

This example collects PI and PD records. The values of the variables in the formula for estimating the number of files are shown below. For the meanings of the variables, see (a) Formulas.
m = 1
n = 1
A11 to A1m = 3 (days)
A21 to A2m = 3 (days)
A31 to A3m = 1 (weeks)
A41 to A4m = 1 (weeks)
A51 to A5m = 1 (months)
B1 to Bn = 10 (days)
The formula is as follows:
N = 20 + 2 x {
        [3(PI minute-by-minute) + 1]  +
        [3(PI minute-by-minute) + 1]  +
        [1(PI minute-by-minute) + 1]  +
        [1(PI minute-by-minute) + 1]  +
        [1(PI minute-by-minute) + 1]  +
        [11 x 1]  +
        [10(PD minute-by-minute) + 1]
       }
= 20 + 2 x {4 + 4 + 2 + 2 + 2 + 11 + 11} = 92
n Number of directories

This example collects PI and PD records. The values of the variables in the formula for estimating the number of directories are shown below. For the meanings of the variables, see (a) Formulas.
A1max = 3 (days)
A2max = 3 (days)
A3max = 1 (week)
A4max = 1 (week)
A5max = 1 (month)
Bmax = 10 (days)
The formula is as follows:
N = 25 + 2 x (3 + 3 + 1 + 1 + 1 + 11 + 10) = 85
n Number of files opened by the Store service

This example collects PI and PD records. The values of the variables m and n in the formula for estimating the number of files that the Store service is to open are shown below. For the meanings of the variables, see (a) Formulas.
m = 1
n = 1
The formula is as follows:
N = 20 + 2 x (6 x 1 + 1) = 34
(c) Disk space requirement for operation in a cluster system

The disk space requirement for operation in a cluster system is the same as for non-cluster system operation. For details, see (a) Formulas.

Record type	Summary type	Variable
a	b	c
PD	--	1,440	g/60	Retention period (days)
PI	Minute	1,440	1 + (g - 1)/60^#	Retention period (days)
Hour	24	1 + (g - 1)/3,600^#	Retention period (days)
Day	7	1 + (g - 1)/86,400^#	Retention period (weeks)
Week	1	1 + (g - 1)/604,800^#	Retention period (weeks)
Month	1	1 + (g - 1)/2,592,000^#	Retention period (months)
Year	1	1 + (g - 1)/31,622,400^#	Retention period (years)

Record type	Data type	Retention period
PD	--	7 days
PI	Minute-by-minute	1 day
Hourly	7 days
Daily	54 weeks
Weekly	54 weeks
Monthly	1 year
Yearly	Unlimited

Record	How to estimate the number of instances	Checking required
PD	1 (because this is a single-instance record)	--
PI	1 (because this is a single-instance record)	--
PI_CPU	Number of processors on the monitored host + 1 (_Total).	Y
PI_LDSK	Number of logical disks on the monitored host + 1 (_Total).	Y
PI_NET	Number of network interfaces on the monitored host + 1 (_Total).	Y
PI_PDSK	Number of physical disks allocated to the monitored host + 1 (_Total).	Y

Record	Description
PI_CPU	Check the number displayed for the instances of the following performance object (including _Total): Processor
PI_LDSK	Check the number displayed for the instances of the following performance object (including _Total): Logical disk
PI_NET	Check the number displayed for the instances of the following performance object + 1 (including _Total): Network interface
PI_PDSK	Check the number displayed for the instances of the following performance object (including _Total): Physical disk

Record	Commands	Description
HP-UX	Solaris	AIX	Linux
PI_CPU	sar -Muw 1 1	mpstat -p 1 2	sar -u -P ALL 1 1	mpstat -P	The value returned by the command is the number of CPUs + 1.
PI_LDSK	df -lk	df -lk	df -k	df -lkP	The value returned by the command is the number of file systems + 1.
PI_NET	netstat -in	netstat -in	netstat -in	netstat -in	The value returned by the command is the number of interfaces + 1.
PI_PDSK	iostat	iostat -x 1 1	sar -d 1 1	iostat -xkd 1 1	The value returned by the command is the number of disk devices + 1.

[Contents][Back][Next]

[Trademarks]