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

[Contents][Glossary][Index][Back][Next]

Appendix A.2 Estimating disk space requirements

Disk space requirements vary according to the number of records for which performance data is collected.

This section explains how to estimate the disk space requirements of PFM - Agent for Platform.

Organization of this subsection
(1) Disk space requirements of the entire system
(2) Disk space requirements of the Store database (Store 1.0)
(3) Disk space requirements of the Store database (Store 2.0)

(1) Disk space requirements of the entire system

Table A-2 Disk space requirements of the entire system

State of PFM - Agent for Platform Disk space requirements (units: MB)
During installation 20
Operation in the initial state 100
Other operating states 50 + W

Legend:
W: Disk space requirements of the Store database

Note:
When installing a patch, you must estimate disk space requirements separately for each patch (the maximum size being the same as that during installation).

The disk space requirements of the Store database are computed by adding together the disk space requirements for PI- and PD-type records.

For the formula for estimating the disk space requirements by record type, see below.

(2) Disk space requirements of the Store database (Store 1.0)

The following explains the disk space requirements of the Store database (Store 1.0).

(a) Formula for estimating disk space requirements

In the Store database, records of the same type are stored in the same file. The table below shows the disk space requirements of the Store database (Store 1.0) by record type.

Table A-3 Disk space requirements of the Store database by record type

Record type Disk space requirements estimation formula (units: bytes)
PI record type (X1 + ..... + Xa + 3,500 x a)
PD record type (Y1 + ..... + Yb + 700 x b)

Legend:
X: Disk space requirements of each record for which historical data is collected as PI-type records

X is computed as follows:
X = {e x f + (d + 1,900) x {(e x f)/(65,250 - d) + 1}#1} x g x 1.5
Y: Disk space requirements of each record for which historical data is collected as PD-type records

Y is computed as follows:
Y ={e x h + (d + 1,900) x {(e x f)/(65,250 - d) + 1}#1 x (h/f)#2} x 1.5
a: Number of records for which historical data is collected as PI-type records
b: Number of records for which historical data is collected as PD-type records
d: Size of the fixed portion of each record for which historical data is collected#3
e: Size of the variable portion of each record for which historical data is collected#3
f: Number of instances in the record for which historical data is collected (1 for a single-instance record)#6
g: Number of saved records for which historical data is collected (the number of instances is not considered)#4
h: Upper limit in the number of saved records for which historical data is collected#5

#1
Truncate the computation result of {(e x f)/(65,250 - d) + 1} at the decimal point.

#2
Truncate the computation result of (h/f) at the decimal point.

#3
For details about the fixed and variable portions of each record, see Summarization rules for each record in 5. Records.

#4
For PI-type records, the collected data is automatically summarized according to certain units (hour, day, week, month, and year). Therefore, when estimating the number of records saved, you must take into account the numbers of records saved for hour, day, week, month, and year. The table below shows the default retention period and the default number of records saved.

Table A-4 Default retention period and the default number of records saved

Data type Retention period Number of records saved
(when the collection interval is 1 minute)
Minute by minute 1 day 1,440
Hourly 7 days 168
Daily 1 year 366
Weekly 1 year 52
Monthly 1 year 12
Yearly No limit (number-of-years-collected) x 1

#5
For details about the number of records saved, see E.1 Agent Store service properties.

#6
You can estimate the number of instances for the records listed below as the number of instances of the performance objects that correspond to records, by using one of the Windows administration tools, System Monitor or Performance Logs and Alerts, in the Performance console.

Table A-5 Number of instances of performance objects that correspond to records

Record ID Performance objects that correspond to records Remarks
PD Process Number of instances
PD_PAGF Paging File Number of instances
PD_PDI Process Number of instances
PD_PEND Process Number of instances that end during the collection interval
PI _total instances of Redirector, Server, Cache, Memory, Objects, System, and Processor 1 because this is a single-instance record
PI_AD lsass instances of NTDS, DNS, and Database 1 because this is a single-instance record
PI_BRSR Browser 1 because this is a single-instance record
PI_ICM6 ICMPv6 1 because this is a single-instance record
PI_ICMP ICMP 1 because this is a single-instance record
PI_IP IP or IPv4 1 because this is a single-instance record
PI_IP6 IP or IPv6 1 because this is a single-instance record
PI_LOGD LogicalDisk Number of instances
PI_NBT NBT Connection Number of instances
PI_NETI Network Interface Number of instances
PI_PCSR Processor Number of instances
PI_PHYD PhysicalDisk Number of instances
PI_SVRQ Server Work Queues Number of instances
PI_TCP TCP or TCPv4 1 because this is a single-instance record
PI_TCP6 TCP or TCPv6 1 because this is a single-instance record
PI_UDP UDP or UDPv4 1 because this is a single-instance record
PI_UDP6 UDP or UDPv6 1 because this is a single-instance record
PI_WINS WINS Server 1 because this is a single-instance record
The number of instances for the records listed below can be estimated from the number of specified user records by using a Windows tool, the collection data addition utility, or the function for collecting user-created data.

Table A-6 Number of instances that can be checked by using a Windows tool, the collection data addition utility, or Windows tool, the function for collecting user-created data

Record ID Checked with Remarks
PD_APP Administration Tools - Services Number of items that are set
PD_DEV System Tools - System Information - Software Environment - Drivers Number of drivers registered
PD_ELOG Administration Tools - Event Viewer Number of logs output during the collection interval
PD_GEND Collection data addition utility Number of user records that are set
PD_SVC Administration Tools - Services Number of services registered
PD_UPD Function for collecting user-created data Number of items that are set
PD_UPDB Function for collecting user-created data Number of items that are set
PI_GENI Collection data addition utility Number of user records that are set
PI_UPI Function for collecting user-created data Number of items that are set
PI_UPIB Function for collecting user-created data Number of items that are set
PI_WGRP Collection data addition utility Number of user records that are set

(b) Examples of estimating disk space requirements

Example 1:
This example estimates the disk space requirements of the PI-type Processor Overview (PI_PCSR) records when PI-type records other than PI_PCSR are not specified to be saved. The variables a to g in Table A-3 have the following values:
a = 1
d = 681
e = 532
f = 2
g = 2,039

How to obtain the value of f
See the number of instances of performance objects that correspond to records in Table A-5. In this example, the value of f is 2.

How to obtain the value of g
For PI_LOGD, when the collection interval is one minute, the yearly collection period is 1 year, and the retention period is set as shown in #4 of Table A-3, g is computed as follows:
1,440 + 168 + 366 + 52 + 12 + 1 = 2,039 records
g = 2,039

Formula for computing X:
X = {e x f + (d + 1,900) x {e x f/(65,250 - d) + 1}} x g x 1.5
X = {532 x 2 + (681 + 1,900) x {532 x 2/(65,250-681) + 1}} x 2,039 x 1.5
 = {1,064 + (2,581) x {1,064/64,569 + 1}} x 3,058.5
 = {1,064 + 2,581 x 1} x 3,058.5
 = 3,645 x 3,058.5
 = 11,148,232 (bytes)
 = Approximately 10.6 (MB)

Example 2:
This example estimates the disk space requirements of the PD-type Process Detail (PD) records when the OS is Windows and PD-type records other than PD are not specified to be saved. The variables in Table A-3 have the following values:
b = 1
d = 681
e = 472
f = 3
h = 2,232

How to obtain the value of f
See the number of instances of performance objects that correspond to records in Table A-5. In this example, the value of f is 3.

How to obtain the value of h
The default number of records saved for the retention period is 10,000.
When you want to set the collection interval to 3,600 seconds and retain data for one month, the value of h is obtained as follows:
h = 24 records (for one day) x 31 (one month) x 3 (value of f) = 2,232 records
h = 2,232 records

Formula for computing Y:
Y = {e x h + (d + 1,900) x {(e x f)/(65,250 - d) + 1} x (h/f)} x 1.5 
Y = {472 x 2,232 + (681 + 1,900) x {(472 x 3)/(65,250 - 681) + 1} x (2,232/3)} x 1.5
  = {472 x 2,232 + 2,581 x {1,416/64,569 + 1} x 744} x 1.5
  = {1,053,504 + 2,581 x 1 x 744} x 1.5
  = {1,053,504 + 1,920,264} x 1.5
  = 2,973,768 x 1.5
  = 4,460,652 (bytes)
  = Approximately 4 (MB)

(3) Disk space requirements of the Store database (Store 2.0)

The following explains the disk space requirements of the Store database (Store 2.0).

(a) Formulas for estimating disk space requirements, number of files, and number of directories

The following explains how to estimate the disk space requirements, the number of files, and the number of directories.

n Disk space requirements

The disk space requirements of the Store database are the sum of the disk space requirements for each disk type. For the PI record type, the disk space requirements are also the sum of the disk space requirements for each summarization category.

Note:
When performance data is stored in the Store database, several fields are added. Because the added fields are included in the disk space requirements, there is no need to make a new estimate. The fields added to all records are the same as the fields in Store 1.0. For details, see (2) Disk space requirements of the Store database (Store 1.0).

Formula for estimating disk space requirement X by record type (units: bytes)
X = {(e + 2) x f + (d + 60) x {((e + 2) x f)/(65,250 - d) + 1}#1} x a/b x (c + 1) x 1.1
a: This value varies depending on the record type and the summarization category. See Table A-7.
b: This value varies depending on the record type and the summarization category. See Table A-7#2
c: Historical-data retention period setting#3. The unit to be specified varies depending on the record type and the summarization category. For details about the units, see Table A-7.
d: Size of the fixed portion of each record for which historical data is collected#4
e: Size of the variable portion of each record for which historical data is collected#4
f: Number of instances in the record for which historical data is collected (1 for a single-instance record)#5. When the number of instances is 2 or greater, the number is rounded to the nearest multiple of 4. For example, if the number of instances is 2, the value of f is 4. If the number of instances is 13, the value of f is 16. If the number of instances is 1, the value of f is 1.

Table A-7 Values set in a, b, and c

Record type Summarization category a b c
PI Minute 1,440 1 + (g - 1)/60#2 Retention period (units: day)
Hour 24 1 + (g - 1)/3,600#2 Retention period (units: day)
Day 7 1 + (g - 1)/86,400#2 Retention period (units: week)
Week 1 1 + (g - 1)/604,800#2 Retention period (units: week)
Month 1 1 + (g - 1)/2,592,000#2 Retention period (units: month)
Year 1 1 + (g - 1)/31,622,400#2 Retention period (units: year)
PD -- 1,440 g/60 Retention period (units: day)
PL -- 1,440 g/60 Retention period (units: day)

Legend:
g: Setting of the historical data collection interval (units: second)
--: Not applicable

#1
Truncate the computation result of {((e + 2) x f)/(65,250 - d) + 1} at the decimal point.

#2
For PI-type records, truncate the computation result of b at the decimal point.

#3
Tables A-8 and A-9 show the default retention periods for Store 2.0.

#4
For details about the size of the fixed portion and variable portion of each record, see the description about record size in 5. Records.

#5
For details about the number of instances for each record, see (2) Disk space requirements of the Store database (Store 1.0)

Table A-8 Default retention period for the PI record type

Data type Retention period
By the minute 1 day
Hourly 7 days
Daily 54 weeks
Weekly 54 weeks
Monthly 12 months
Yearly No limit

Table A-9 Default retention period for the PD record type

Record name Retention period (units: day)
PD 2
PD_PDI 2
PD_PEND 2
PD_PAGF 2
PD_SVC 2
PD_GEND 10
PD_DEV 2
PD_ELOG 31
PD_UPD 10
PD_UPDB 10
PD_APP 10

n Number of files

The following shows the formula for estimating the number of files N that are created in 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) +
         (C1 + C2 + ... + Co + o)
         )

m: Number of records collected as PI-type records

n: Number of records collected as PD-type records

o: Number of records collected as PL-type records

A11 to A1m: Minute-by-minute record retention period setting for each PI-type record (units: day)

A21 to A2m: Hourly record retention period setting for each PI-type record (units: day)

A31 to A3m: Daily record retention period setting for each PI-type record (units: week)

A41 to A4m: Weekly record retention period setting for each PI-type record (units: week)

A51 to A5m: Monthly record retention period setting for each PI-type record (units: month)

B1 to Bn: Retention period setting for each PD-type record (units: day)

C1 to Co: Retention period setting for each PL-type record (units: day)

n Number of directories

The following shows the formula for estimating the number of directories N that are created in the Store database.

N = 25 + 2 x ((A1max) + (A2max) + (A3max) + (A4max) + (A5max) + 11 + (Bmax) + (Cmax))

A1max: Maximum retention period setting for the data of the records which are collected as PI-type records and whose summarization category is Minute (units: day)

A2max: Maximum retention period setting for the data of the records which are collected as PI-type records and whose summarization category is Hour (units: day)

A3max: Maximum retention period setting for the data of the records which are collected as PI-type records and whose summarization category is Day (units: week)

A4max: Maximum retention period setting for the data of the records which are collected as PI-type records and whose summarization category is Week (units: week)

A5max: Maximum retention period setting for the data of the records which are collected as PI-type records and whose summarization category is Month (units: month)

Bmax: Maximum retention period setting for each PD-type record (units: day)

Cmax: Maximum retention period setting for each PL-type record (units: day)

n Number of files that the Store service opens

The following shows the formula for estimating the number of files N that the Store service opens:

N = 20 + 2 x (6 x l + m + n)

l: Number of records collected as PI-type records

m: Number of records collected as PD-type records

n: Number of records collected as PL-type records

(b) Examples of estimating disk space requirements

The following provides an example of estimating the disk space requirements of the Store database (Store 2.0) of PFM - Agent for Platform.

n Disk space requirements

This example shows how to estimate the disk space requirements when collection of the PI_LOGD and PD records is specified.

The following explains how to estimate the disk space requirements of the PI_LOGD record. The variables shown in the formula in (a) Formula for estimating disk space requirements, number of files, and number of directories have the following values:

d = 681 (bytes)
e = 700 (bytes)
f = 4
g = 60 (seconds)

The following shows how to calculate the disk space requirements by record type, such as a record collected by the minute or hourly.

Minute-by-minute record
The variables have the following values:
a = 1,440
b = 1 + (60-1)/60
  = 1.98... 
  = 1 (Digits after the decimal point are truncated.)
c = 3 (days)
The formula is as follows:
X (minute) = {(700 + 2) x 4 + (681 + 60) x {((700 + 2) x 4)/(65,250 - 681) + 1}} x 1,440/1 x (3 + 1) x 1.1
       = {2,808 + 741 x 1} x 6,336
       = 3,549 x 6,336
       = 22,486,464 (bytes) = Approximately 22 (MB)

Hourly record
The variables have the following values:
a = 24
b = 1 + (60-1)/3,600
  = 1.01...
  = 1 (Digits after the decimal point are truncated.)
c = 3 (day)
The formula is as follows:
X (hour) = {(700 + 2) x 4 + (681 + 60) x {((700 + 2) x 4)/(65,250-681) + 1}} x 24/1 x (3 + 1) x 1.1
       = {2,808 + 741 x 1} x 105.6
       = 3,549 x 105.6
       = 374,774.4 (bytes) = Approximately 0.4 (MB)

Daily record
The variables have the following values:
a = 7
b = 1 + (60-1)/86,400
  = 1.00...
  = 1 (Digits after the decimal point are truncated.)
c = 1 (week)
The formula is as follows:
X (day) = {(700 + 2) x 4 + (681 + 60) x {((700 + 2) x 4)/(65,250-681) + 1}} x 7/1 x (1 + 1) x 1.1
       = {2,808 + 741 x 1} x 15.4
       = 3,549 x 15.4
       = 54,654.6 (bytes) = Approximately 0.05 (MB)

Weekly record
The variables have the following values:
a = 1
b = 1 + (60-1)/604,800
  = 1.00...
  = 1 (Digits after the decimal point are truncated.)
c = 1 (week)
The formula is as follows:
X (week) = {(700 + 2) x 4 + (681 + 60) x {((700 + 2) x 4)/(65,250-681) + 1}} x 1/1 x (1 + 1) x 1.1
       = {2,808 + 741 x 1} x 2.2
       = 3,549 x 2.2
       = 7,807.8 (bytes) = Approximately 0.008 (MB)

Monthly record
The variables have the following values:
a = 1
b = 1 + (60-1)/2,592,000
  = 1.00...
  = 1 (Digits after the decimal point are truncated.)
c = 1 (month)
The formula is as follows:
X (month) = {(700 + 2) x 4 + (681 + 60) x {((700 + 2) x 4)/(65,250-681) + 1}} x 1/1 x (1 + 1) x 1.1
       = {2,808 + 741 x 1} x 2.2
       = 3,549 x 2.2
       = 7,807.8 (bytes) = Approximately 0.008 (MB)

Yearly record
The variables have the following values:
a = 1
b = 1 + (60-1)/31,622,400
  = 1.00...
  = 1 (Digits after the decimal point are truncated.)
c = 10 (fixed)
The formula is as follows:
X (year) = {(700 + 2) x 4 + (681 + 60) x {((700 + 2) x 4)/(65,250-681) + 1}} x 1/1 x (10 + 1) x 1.1
       = {2,808 + 741 x 1} x 12.1
       = 3,549 x 12.1
       = 42,942.9 (bytes) = Approximately 0.04 (MB)

From the above calculations, the disk space requirements of the PI_LOGD record are obtained as follows:

X (total) = X (minute) + X (hour) + X (day) + X (week) + X (month) + X (year)
          = 22.506 (MB)
          = Approximately 23 (MB)

The following explains how to estimate the disk space requirements of the PD record.

The variables shave the following values:

a = 1,400
b = 60/60 = 1
c = 10 (day)
d = 681 (bytes)
e = 472 (bytes)
f = 12
g = 60 (seconds)

The formula is as follows:

X = {(472 + 2) x 12 + (681 + 60) x {((472 + 2) x 12)/(65,250 - 681) + 1} x 1,400/1 x (10 + 1) x 1.1
  = {5,688 + 741 x 1} x 17,424
  = 6,429 x 17,424
  = 112,018,896 (bytes) = Approximately 112 (MB)

The disk space requirements are therefore PI_LOGD + PD = 135 MB.

n Number of files

This example shows how to estimate the number of files when the PI, PI_LOGD, PD, and PD_PDI records are to be collected. The variables shown in the formula in (a) Formula for estimating disk space requirements, number of files, and number of directories have the following values:

m = 2
n = 2
o = None
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)
C1 to Co = Not required

The following shows the formula for estimating the number of files N that are created in 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) +
         (C1 + C2 + ... + Co + o)
         )
  = 20 + 2 x {
         [3 (for PI) + 3 (for PI_LOGD) + 2] +
         [3 (for PI) + 3 (for PI_LOGD) + 2] +
         [1 (for PI) + 1 (for PI_LOGD) + 2] +
         [1 (for PI) + 1 (for PI_LOGD) + 2] +
         [1 (for PI) + 1 (for PI_LOGD) + 2] +
         [11 x 2] + 
         [10 (for PD) + 10 (for PD_PDI) + 2]
         }
  = 20 + 2 x {8 + 8 + 4 + 4 + 4 + 22 + 22}
  = 20 + 2 x 72 = 20 + 144 = 164

n Number of directories

This example shows how to estimate the number of directories when the PI, PI_LOGD, PD, and PD_PDI records are to be collected.

The variables shown in the formula for the number of directories in (a) Formula for estimating disk space requirements, number of files, and number of directories have the following values:

A1max = 3 (days) (Principle: When the setting for PI is 2 days and the setting for PI_LOGD is 3 days, 3 days is used.)
A2max = 3 (days)
A3max = 1 (weeks)
A4max = 1 (weeks)
A5max = 1 (months)
Bmax = 10 (days) (Principle: When the setting for PD_PDI is 8 days and the setting for PI is 10 days, 10 days is used.)
Cmax = None

The following shows the formula for estimating the number of directories N that are created in the Store database:

N = 25 + 2 x ((A1max) + (A2max) + (A3max) + (A4max) + (A5max) + 11 + (Bmax) + (Cmax))
  = 25 + 2 x (3 + 3 + 1 + 1 + 1 + 11 + 10 + 0) = 25 + 2 x 30 = 85

n Number of files that the Store service opens

This example shows how to estimate the number of files that the Store service opens when the PI, PI_LOGD, PD, and PD_PDI records are to be collected.

The variables shown in the estimation formula in (a) Formula for estimating disk space requirements, number of files, and number of directories have the following values:

l = 2
m = 2
n = 0

The following shows the formula for estimating the number of files N that the Store service opens:

N = 20 + 2 x (6 x l + m + n)
  = 20 + 2 x (6 x 2 + 2 + 0) = 48

[Contents][Back][Next]

[Trademarks]

All Rights Reserved. Copyright (C) 2009, Hitachi, Ltd.