Fdm 92000 Architecture Analysis Guide

7/31/2019 Fdm 92000 Architecture Analysis Guide

1/40

Hyperion System 9 Financial

Data Quality Management

Architecture Analysis Guide

Hyperion FDM 9.2.0


2/40


3/40

i

Table of Contents

PurPose. 1

Audience. 1

document.structure. 1

Overview. 1

TesT.Lab.envirOnmenT. 1

DaTa.viewing.COnCurrenCy.TesT. 1

DaTa.LOaDing.by.DaTabase.Type.TesT. 1

DaTa.LOaD.perfOrmanCe.Over.Time.TesT. 1

DaTa.LOaDing.COnCurrenCy.TesT. 1

sysTem.reCOmmenDaTiOns 1

overview. 2

AdditionAl.Architecture.considerAtions. 2

test.lAb.environment. 2

server.hArdwAre. 3

webserver.(DeLLTesT2). 3

appLiCaTiOn.server.(DeLLTesT1). 3



DaTa.server.(DeLLTesT5). 3

DaTa.server.(DeLLTesT6). 3

dAtA.viewing.concurrency.test.(test.#1). 4

meThODOLOgy. 4

resuLTs. 4

dAtA.loAding.by.dAtAbAse.tyPe.test.(test.#2). 6

TesT.meThODOLOgy 6

TesT.resuLTs. 6dAtA.loAd.PerformAnce.over.time.(test.#3). 9

TesT.meThODOLOgy 9

TesT.resuLTs. 9

dAtA.loAding.concurrency.test.(test.#4). 12

TesT.meThODOLOgy 12

TesT.resuLTs. 12

system.recommendAtions. 14

DaTabase.server. 14

appLiCaTiOn.server. 14

web.server. 14

appLiCaTiOn.seTTings. 15

fiLe.sharing.pOrTs. 15

firewaLL./.DCOm.seTTings. 16

TCp.Keep.aLive.seTTings. 16

OraCLe.iniTiaLizaTiOn.parameTers 17


4/40

ii


5/40

PurposeThe purpose of this document is to provide assistance in determining the required infrastructure and optimaldeployment architecture for a given number of concurrent Hyperion System 9 Financial Data Quality Management(Hyperion FDM) users.

AudienceThis guide is intended for IT system designers and business decision makers who are responsible for determiningapplication scalability and functionality.

Document StructureThis document contains the following major topics of information:

OverviewIntroduces the product architecture considerations.

Test Lab EnvironmentDetails the Hyperion test lab environment that was used during the stress testing detailed in this document.

Data Viewing Concurrency TestProvides testing results for a simulation that simulates a high volume of concurrent users viewing data in theHyperion FDM application and the resulting impact on the web server.

Data Loading by Database Type Test

Provides testing results for a simulation that simulates loading increasing le sizes for all supported databasesand the resulting impact on performance.

Data Load Performance Over Time TestProvides testing results from a simulation that simulates high volume loads for several concurrent months andthe resulting impact on database size and processing time.

Data Loading Concurrency TestProvides testing results for a simulation that simulates an increasing number of concurrent users loading dataand the resulting impact on the application and data servers.

System RecommendationsProvides users with recommendations for designing and optimizing their own Hyperion FDM application.


6/40

Hyperion FDM Architecture Analysis Guide

OverviewYou can use the test results described in this document to assist in designing and optimizing your Hyperion System

9 Financial Data Quality Management hardware architecture.

Additional Architecture ConsiderationsHyperion FDM can be deployed over multiple architectural environments. The most common architecture consistsof one web server, one database server, and one or more application servers. The optimal architecture dependson the types of processing conducted in your Hyperion FDM application. As different types of processing are doneon each system component, it is necessary to determine what types of processing will be used most extensivelyin your system to determine the optimal conguration. For example, you may need more web servers if you have

a large volume of users viewing data, or you may need to upgrade your database server if you use extensivewild cards in your mapping tables. Below are the major Hyperion FDM processes you will need to consider whendesigning your system architecture.

Types of System Processes in Hyperion FDM:

Database Server Intensive Tasks Deleting previously imported Hyperion FDM data prior to a load,executing bulk inserts during a load, mapping calculations after a load, report querying

Application Server Intensive Tasks Parsing and reading source les during the load (import text

process), serving Hyperion FDM reports.

Hyperion FDM Web Server Intensive Tasks Serving web pages

Test Lab EnvironmentThe test environment architecture consists of a single web server, three application servers, and one data server.

All servers are dual processor machines with a minimum of 2 GB of RAM.


7/40

Server HardwareNote: All servers were Dell Power Edge 2650

Webserver (DellTest2)Processor 2.8 GHz Intel Xeon Dual processor

OS Win 2003 Server

Memory 2 GB RAM

HDD 2 x 30 GB SCSI HDD

Application Server (Delltest1)

OS Win 2003 Server

Processor 2.8 GHz Intel Xeon Dual processor

Memory 2 GB RAM



OS Win 2003 Server


Memory 2 GB RAM



OS Win 2003 Server


Memory 2 GB RAM


Data Server (Delltest5)

OS Win 2003 Server


Memory 4 GB RAM


Database SQL 2000, Oracle 9i

Data Server (Delltest6)OS Win 2003 Server


Memory 4 GB RAM


Database SQL 2005, Oracle 10g


8/40


Data Viewing Concurrency Test (Test #1)

Methodology

1. This testing method is focused on web server processing for an increasing numbers of concurrent users.

2. The test demonstrates the Web Servers ability to interpret and send a high volume of requests and theApplication/Data Servers ability to respond to these requests.

3. The Browser Requests for this test focus on data viewing/retrieval and not on data updating/inserting.

4. The Hyperion FDM Concurrency test comprises the following operations:

a. Log into application

b. Select Maps link

c. Remain logged onto the application

ResultsTest #1 Concurrent Data Viewing Open 500 data consumer connections simultaneously.

This test produced 500 connections during the entire test run. The rst login occurred at 4:21:20 PM and the last

login occurred at 4:25:08 PM. The entire test took 5 minutes and 52 seconds to complete.

Web Server Connection Results

Number User Connected over Time

0

5

10

15

20

25

30

520

35

50

65

80

95

110

125

140

155

170

185

200

215

230

Time Interval (Sec)

NumberofUsers

Number User Connected

Interpreting the Results

Although the test is meant to represent 500 concurrent users, not all of the users are beginning the loginprocess at the same instant. A delay of approximately .5 - 2 seconds was randomly inserted between requestssent to the Hyperion FDM application. This is meant to more accurately simulate real world concurrency. Thesingle web server was the bottleneck in this testing scenario. The Web Server processor averaged 90%utilization during the test runs. The addition of another web server would have improved overall performance.

Performance could also be improved by deploying a more powerful web server.


9/40

Microsoft Application Center Test Results

Test Name: Web Server Concurrency Test Run

Test Started: 3/9/2006 4:21:14 PM

Test Duration: 00:00:05:52Test Iterations: 500

Properties

Simultaneous browser connections: 500

Test duration: 00:00:05:52

Summary

Total number of requests: 42,500

Total number of connections: 42,500

Average requests per second: 120.74

Average time to rst byte (msecs): 1,110.59

Average time to last byte (msecs): 1,111.21

Average time to last byte per iteration (msecs): 94,452.44

Errors Counts

HTTP: 0

DNS: 0

Socket: 0

Additional Network Statistics

Average bandwidth (bytes/sec): 879,063.63

Number of bytes sent (bytes): 21,398,412

Number of bytes received (bytes): 288,031,986

Average rate of sent bytes (bytes/sec): 60,790.94

Average rate of received bytes (bytes/sec): 818,272.69


10/40


Data Loading by Database Type Test (Test #2)

Test Methodology

1. This testing method focuses on application and data server update/insert processing for increasing lesize loads.

2. This test demonstrates the performance of each of the databases supported as le sizes increase.

3. The Hyperion FDM Import Workow task comprises the following operations:

a. Login to application

b. Delete existing data records

c. Delete existing data archive records and les

d. Read and parses the specied import text le

e. Write the cleansed import le to database

f. Add new le to Hyperion FDM data archive

g. Execute Bulk Load

h. Execute Hyperion FDM Mapping Rules

i. Update Workow Status

j. Logoff application

4. This test process is repeated using SQL 200, SQL 2005, Oracle 9i & Oracle 10g.

Test ResultsData Loading by Database Type Files ranging from 5000 records to 2 million records were loaded into eachof the supported databases. Load time for the 1 million record loads were as follows: SQL 2005 6.18 minutes,SQL 2000 6.90 minutes, Oracle 10g 11.15 minutes, Oracle 9i 13.43 minutes.

Data Load Duration Results

SQL 2005 SQL 2000 Oracle 10g Oracle 9i

200K 74 88 123 127

500K 174 209 299 343

1000K 371 414 669 806

2000K 808 909 1693 3122

The rst graph (next page) shows database import times as le sizes grow. For le sizes 200,000 records

and above both SQL databases perform signicantly better than the Oracle databases. As the le sizes grow

the performance differences become more evident. For le sizes below 200,000 records the performance

differences between all database types was minimal.

The second graph depicts the breakdown of the individual steps of the import process. Although most of theindividual steps have comparable times for Oracle & SQL, Oracle mapping takes signicantly longer than

the same mapping process for SQL. This is primarily due to the fact that Oracle does not support updates onjoined tables. This action requires Oracle to perform cursor updates which take signicantly longer.


11/40

Import Process Breakdown for One Million Records

(seconds)

59

37

69

40

132

124

135

37

36

51

18

34

21

46

103

93

313

319

49

60

148

106

18

15

161

23

0 100 200 300 400 500 600 700 800

SQL-2000

SQL-2005

Oracle-9i

Oracle-10g

DatabaseType

Import Process Duration (seconds)

Delete Data

Import

File Archive

Create WT Index

Map

Post WorkData

Drop Work Table


12/40



SQL databases run signicantly faster than Oracle for high-volume les. Hyperion FDM has been tuned to

optimize performance for data loading, this results in data scanning during loads for Oracle. Performance

testing has resulted in this being the optimal method for Oracle as indexing the required tables resulted in

decreased performance especially as the Oracle database grows in size.


13/40

9

Data Load Performance over Time (Test #3)

Test Methodology

1. This testing method focuses on overall processing time and database size for a typical high-volumemonth-end scenario with a small number of users loading a high volume of data.

2. The test was run 12 times (simulating 12 months) to show the database sizing as well as any possible

performance degradation as database size increases.


a. Log into application

b. Read and parses the specied Import text le

c. Write the cleansed import le to database

d. Add new le to Hyperion FDM data archive

e. Execute Bulk Load

f. Execute Hyperion FDM Mapping Rulesg. Update Workow Status

h. Logoff application

4. The test was run once for SQL 2005 and once for Oracle 10g.

Test ResultsData Load Performance over Time Loading one million records each month for 12 months. This test wasrun for both SQL 2005 & Oracle 10g. To enhance performance, separate hard drives on the data base serverwere used to store the mdf/log les for SQL and the table spaces for Oracle.

Data Load Duration & Sizing Results

File Structure Size

(GB)Database Size (GB) Load Time

(hh:mm:ss)

Archive Logs SQL 2005 Oracle 10g SQL 2005 Oracle 10g

Month1 .11 .02 .59 .30 00:05:56 00:08:21

Month2 .23 .05 1.16 .60 00:05:49 00:08:48

Month3 .34 .07 1.74 .90 00:05:57 00:09:40

Month4 .46 .10 2.11 1.19 00:05:53 00:08:46

Month5 .57 .12 2.49 1.49 00:05:51 00:09:11

Month6 .69 .14 2.46 1.79 00:05:48 00:09:07

Month7 .80 .17 2.87 2.09 00:05:53 00:10:52

Month8 .92 .19 3.63 2.39 00:05:57 00:11:58

Month9 1.03 .21 3.54 2.68 00:05:47 00:12:08Month10 1.14 .24 3.91 2.98 00:05:52 00:11:39

Month11 1.26 .26 4.28 3.28 00:06:29 00:13:02

Month12 1.37 .29 4.66 3.58 00:06:37 00:12:30


14/40


0


(SQL 2005)

0 100 200 300 400 500 600 700 800

Mth1

Mth2

Mth3

Mth4

Mth5

Mth6

Mth7

Mth8

Mth9

Mth10

Mth11Mth12

Month

Time (Seconds)

Import: Text File

Import: Create WorkTable Indexes

Import: Process Map

Import: Delete

Import: Post WorkData

Check Trial Balance

Import: Drop Work

Table


(Oracle 10g)

0 100 200 300 400 500 600 700 800

Mth1

Mth2

Mth3

Mth4

Mth5

Mth6

Mth7

Mth8

Mth9

Mth10

Mth11

Mth12

Month

Time (Seconds)

Import: Text File

Import: Create WorkTable Indexes

Import: Process Map

Import: Delete

Import: Post Work Data

Check Trial Balance

Import: Drop WorkTable


15/40


16/40


Data Loading Concurrency Test (Test #4)

Test Methodology

1. This testing method focuses on application and data server update/insert processing for an increasingnumbers of concurrent users.

2. This test demonstrates the products scalability from a data architecture perspective by executing anincreasing number of simultaneous Import workow tasks.


a. Login to application

b. Delete existing data records

c. Delete existing data archive records and les

d. Read and parses the specied import text le

e. Write the cleansed import le to database

f. Add new le to Hyperion FDM data archive

g. Execute Bulk Load

h. Execute Hyperion FDM Mapping Rules

i. Update Workow Status

j. Logoff application

Test ResultsData Load Concurrency Test Concurrent data loading (100, 200, 300 users, loading 6,000 records). Timeto complete the 100 concurrent user load process using three application servers was 1:40 minutes for 100users, 3:16 minutes for 200 users, and 4:43 minutes for 300 users.

The les that were loaded during this test run were 600KB. The load requests were sent with an average delay

of 2 seconds between users to more realistically represent a normal monthly process.

The graph on the next page represents average duration for the main import process tasks. By running the testmultiple times for increasing number of users, we can see the linear increase in duration for the import text task.The time to import the text into the Hyperion FDM database becomes larger as more users are simultaneously

attempting to write records to the database.


17/40

Average Import Process Duration (by Type)

0

50

100

150

200

250

300

100 Users 200 Users 300 Users

Number of Users Loading

Time(Seconds)

Delete

Import Text

Create WT Index

Map

Post WD

Drop WT

Total Import


During this test, the database server was the bottleneck. As can be seen from the graph above, as moreusers were added the items that are database-intensive (Mapping, Deleting, and Indexing) increase linearly.This is due to the database server being overwhelmed with requests to write to the database. This shows thenecessity of ensuring that the database server is properly sized. Adding database processors in this scenario

would not necessarily improve performance as much as using faster hard drives for the database server.

This test was performed on SQL 2005 with the database log and indexing/worktables each set up on separatehard disk drives. In this instance, the recommended database server conguration would be to spread the

database tasks (data storage, logging, and indexing) over multiple data server hard disk drives to improvethe overall database server performance.


18/40


System Recommendations

Database Server

Recommended Hardware

Quad P4 processor

1 GB RAM per 75 concurrent users (2 GB Minimum)

Multiple HDD to spread the processing (especially important with Oracle).

Selecting the appropriate data server hardware is critical for optimal performance for the HyperionFDM application. As can be seen from the Concurrent Data Loading Test, the database server canbecome a bottleneck if not properly sized. Even though you can always add more application servers

and improve performance for selected tasks, doing so can quickly overpower the database server.

Database Sizing

Average total disk space per location = .526 GB per month of storage (based on average le size of

95MB & use of 3 custom dimensions). Each additional dimension increases this on average by .025GB. Note: Average disk size recommendations should be adjusted based on estimated average lesize to be loaded, number of custom dimensions in use, and the number of attribute dimensions in

use.

The database le and log le should be on separate physical drives for SQL. For Oracle each

tablespace (Work, Index, Map Seg, and Data Seg) should be on a separate drive.

For improved performance always use the fastest HDD available.

Document attachments will affect total disk space. All attached documents are archived in theapplication le structure and will occupy the same amount of disk space as the original le.

Exported archive les will also affect total disk space. All exported archives reside in the Archive

Restore directory and will occupy the same amount of disk space as the original archive.

Application Server


Dual P4 processor per 75 concurrent users

1 GB RAM per 75 concurrent users

Web Server


Dual P4 processor per 100 concurrent users

2 GB RAM

.NET Process ConfgurationMaxWorkerThreads should be increased to 100 in the Machine.Cong le

MaxIOThreads should be increased to 100 in the Machine.Cong le

Memory Recycling

By default UpStream WebLink version 8.0.6 and beyond (including Hyperion FDM 9.0.2 andHyperion FDM 9.2.0) use IIS memory recycling to force the Microsoft.Net process to recycle memory


19/40

when the process reached 250MB (Windows 2003) or 25% of Web Server memory (Windows 2000).These values are based on a Web Server running 2 GB of RAM. For servers running 4 Gig of RAMthese settings should be adjusted to 150 - 200MB (Windows 2003) and 15% of memory (Windows2000). The default memory limit is extremely important to adjust when your server has 4 GB or more

of RAM.By default version UpStream WebLink 8.0.6 and beyond (including Hyperion FDM 9.0.2 andHyperion FDM 9.2.0) use aspnet_state service to store session values while running the IIS memoryrecycling process. This service must be running on the Web Server for Hyperion FDM to operatecorrectly.

Please refer to the Hardware and Software Requirements document for detailed information on hardwareand software.

Application Settings

Data Segments

Prior to creating the Hyperion FDM locations, it is possible to update Hyperion FDM data segments in the

Hyperion FDM Conguration options. This represents the number of data tables within the Hyperion FDMapplication. These data tables are shared by the data load locations. To avoid data locking, this should beset to 5-8 concurrent data loaders per segment (e.g., 600 concurrent data loaders in one Hyperion FDMapplication would result in data segments of 75-120). Note: This setting applies for each Hyperion FDMapplication.

File Sharing PortsA working folder is dened for each application created in Hyperion FDM. All Hyperion FDM application servers

and the data server share this folder. Because of this, le sharing must be enabled between all Hyperion FDM

application servers and the data server. The ports required are listed below. The account that the data serverservice runs under must be able to read from this share. In the case of SQL Server the account running the

MSSQLSERVER service must be able to read from this share.

For File sharing enable ports:

Application protocol Protocol Ports

NetBIOS Datagram Service UDP 138

NetBIOS Name Resolution UDP 137

NetBIOS Session Service TCP 139

If NET BIOS is turned off then use:

Application protocol Protocol Ports

SMB TCP 445

Note: The le sharing ports need to be opened on all the application servers and the data server. These

ports do not have to be opened on the Web server.


20/40


Firewall / DCOM Settings

Port 135 must be open on all application servers and web servers to allow for two-way DCOMcommunication.

Unlike most Internet applications which have xed TCP and/or UDP ports, DCOM dynamicallyassignsat run timeone TCP port and one UDP port to each executable process serving DCOMobjects on a computer. Because DCOM (by default) is free to use any port between 1024 and 65535when it dynamically selects a port for an application. Conguring your rewall to leave such a wide

range of ports open would present a potential security hole. Open DCOM port range may be restrictedby changing the following registry key:

HKEY_LOCAL_MACHINE\Software\Microsoft\Rpc\Internet

Details on restricting DCOM ports can be found in the Microsoft knowledge base article:

http://support.microsoft.com/kb/154596

Details on other issues that may arise when going through a rewall can be found in the Microsoft knowledge

base article:

http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dndcom/html/msdn_dcomrewall.asp

TCP Keep Alive SettingsDetermines how often TCP sends keep-alive transmissions. TCP sends keep-alive transmissions to verify thatan idle connection is still active. Hyperion recommends that this setting be reduced to hour. Details onupdating the KeepAlive settings can be found in the Microsoft knowledge base article:

http://www.microsoft.com/resources/documentation/Windows/2000/server/reskit/en-us/Default.asp?url=/resources/documentation/Windows/2000/server/reskit/en-us/regentry/58768.asp


21/40

Oracle Initialization Parameters

Oracle 9i

Below are the Oracle 9i Initialization Parameters used the tests in this document.

NAME VALUE DESCRIPTION

O7_DICTIONARY_ACCESSIBILITY

FALSE Version 7 Dictionary Accessibility Support

active_instance_count Number of active instances in the cluster database

aq_tm_processes 1 Number of AQ Time Managers to start

archive_lag_target 0Maximum number of seconds of redos the standbycould lose

audit_sys_operations FALSE Enable sys auditing

audit_trail NONE Enable system auditing

background_core_dump partial Core Size for Background Processes

background_dump_destE:\oracle\admin\StressDM\bdump

Detached process dump directory

backup_tape_io_slaves FALSE BACKUP Tape I/O slaves

bitmap_merge_area_size 1048576 Maximum memory allow for BITMAP MERGE

blank_trimming FALSE Blank trimming semantics parameter

buffer_pool_keepNumber of database blocks/latches in keep bufferpool

buffer_pool_recycleNumber of database blocks/latches in recycle bufferpool

circuits 555 Max number of circuits

cluster_database FALSE If TRUE startup in cluster database mode

cluster_database_instances

1Number of instances to use for sizing cluster db SGA

structures

cluster_interconnects Interconnects for RAC use

commit_point_strength 1Bias this node has toward not preparing in a two-phase commit

compatible 9.2.0.0.0Database will be completely compatible with thissoftware version

control_le_record_keep_

time7 Control le record keep time in days

control_les

E:\oracle\oradata\StressDM\CONTROL01.

CTL, E:\oracle\oradata\StressDM\CONTROL02.CTL, E:\oracle\oradata\StressDM\CONTROL03.CTL

Control le names list

core_dump_destE:\oracle\admin\StressDM\cdump

Core dump directory


22/40



cpu_count 4 Initial number of CPUs for this instance

create_bitmap_area_size 8388608 Size of create bitmap buffer for bitmap index

cursor_sharing SIMILAR Cursor sharing mode

cursor_space_for_time FALSE Use more memory in order to get faster execution

db_16k_cache_size 0 Size of cache for 16K buffers





db_block_buffers 0 Number of database blocks cached in memory

db_block_checking FALSE Data and index block checking

db_block_checksum TRUE Store checksum in db blocks and check during reads

db_block_size 8192 Size of database block in bytesdb_cache_advice ON Buffer cache sizing advisory

db_cache_size 838860800Size of DEFAULT buffer pool for standard block size

buffers

db_create_le_dest Default database location

db_create_online_log_dest_1

Online log/controlle destination #1





db_create_online_log_dest_4 Online log/controlle destination #4



db_domainDirectory part of global database name stored withCREATE DATABA

db_le_multiblock_read_

count8 DB block to be read each IO

db_le_name_convert Datale name convert patterns and strings for

standby/clone db

db_les 200 Max allowable # db les

db_keep_cache_size 0

Size of KEEP buffer pool for standard block size

buffersdb_name StressDM Database name specied in CREATE DATABASE

db_recycle_cache_size 0Size of RECYCLE buffer pool for standard block size

buffers

db_writer_processes 2Number of background database writer processes tostart


23/40

9


dblink_encrypt_login FALSEEnforce password for distributed login always beencrypted

dbwr_io_slaves 0 DBWR I/O slaves

dg_broker_cong_le1

%ORACLE_HOME%\DATABASE\DR1%ORACLE_SID%.DAT

Data guard broker conguration le #1

dg_broker_cong_le2

%ORACLE_HOME%\DATABASE\DR2%ORACLE_SID%.DAT


dg_broker_start FALSE Start Data Guard broker framework (DMON process)

disk_asynch_io TRUE Use asynch I/O for random access devices

dispatchers

(PROTOCOL=TCP)

(SERVICE=StressDMOracleXDB)

Specications of dispatchers

distributed_lock_timeout 60Number of seconds a distributed transaction waitsfor a lock

dml_locks 2440DML locks - one for each table modied in a

transaction

drs_start FALSE Start DG Broker monitor (DMON process)

enqueue_resources 2660 Resources for enqueues

event Debug event control - default null string

fal_client FAL client

fal_server FAL server list

fast_start_io_target 0 Upper bound on recovery readsfast_start_mttr_target 300 MTTR target of forward crash recovery in seconds

fast_start_parallel_rollback

LOWMax number of parallel recovery slaves that may beused

le_mapping FALSE Enable le mapping

lesystemio_options IO operations on lesystem les

xed_date Fixed SYSDATE value

gc_les_to_locks Mapping between le numbers and global cache

locks (DFS)

global_context_pool_size Global Application Context Pool Size in Bytes

global_names FALSE

Enforce that database links have same name as

remote database

hash_area_size 1048576 Size of in-memory hash work area

hash_join_enabled TRUE Enable/disable hash join

hi_shared_memory_address

0SGA starting address (high order 32-bits on 64-bitplatforms)


24/40


0


hs_autoregister TRUEEnable automatic server DD updates in HS agentself-registration

ile Include le in init.orainstance_groups List of instance group names

instance_name StressDMOracle Instance name supported by the instance

instance_number 0 Instance number

java_max_sessionspace_size

0 Max allowed size in bytes of a Java sessionspace

java_pool_size 8388608 Size in bytes of the Java pool

java_soft_sessionspace_limit

0Warning limit on size in bytes of a Java

sessionspace

job_queue_processes 10 Number of job queue slave processes

large_pool_size 8388608 Size in bytes of the large allocation pool

license_max_sessions 0 Maximum number of non-system user sessionsallowed

license_max_users 0Maximum number of named users that can becreated in the databas

license_sessions_warning

0Warning level for number of non-system usersessions

local_listener Local listener

lock_name_spaceLock name space used for generating lock namesfor standby/clone

lock_sga FALSE Lock entire SGA in physical memory

log_archive_dest Archival destination text string

log_archive_dest_1 Archival destination #1 text stringlog_archive_dest_10 Archival destination #10 text string

log_archive_dest_2 Archival destination #2 text string








log_archive_dest_state_1 enable Archival destination #1 state text stringlog_archive_dest_state_10

enable Archival destination #10 state text string

log_archive_dest_state_2 enable Archival destination #2 state text string




25/40







log_archive_duplex_dest Duplex archival destination text string

log_archive_format ARC%S.%T Archival destination format

log_archive_max_processes

2 Maximum number of active ARCH processes

log_archive_min_succeed_dest

1Minimum number of archive destinations that mustsucceed

log_archive_start FALSE Start archival process on SGA initialization

log_archive_trace 0 Establish archivelog operation tracing level

log_buffer 7024640 Redo circular buffer size

log_checkpoint_interval 0 # redo blocks checkpoint threshold

log_checkpoint_timeout 1800Maximum time interval between checkpoints inseconds

log_checkpoints_to_alert FALSE Log checkpoint begin/end to alert le

log_le_name_convert Logle name convert patterns and strings for

standby/clone db

log_parallelism 1 Number of log buffer strands

logmnr_max_persistent_sessions

1 Maximum number of threads to mine

max_commit_

propagation_delay 700 Max age of new snapshot in .01 seconds

max_dispatchers 5 Max number of dispatchers

max_dump_le_size UNLIMITED Maximum size (blocks) of dump le

max_enabled_roles 30 Max number of roles a user can have enabled

max_rollback_segments 122 Max. number of rollback segments in SGA cache

max_shared_servers 20 Max number of shared servers

mts_circuits 555 Max number of circuits

mts_dispatchers(PROTOCOL=TCP)(SERVICE=StressDMOracleXDB)


mts_listener_address Address(es) of network listener

mts_max_dispatchers 5 Max number of dispatchers

mts_max_servers 20 Max number of shared servers

mts_multiple_listeners FALSE Are multiple listeners enabled?

mts_servers 1 Number of shared servers to start up

mts_service StressDM Service supported by dispatchers


26/40



mts_sessions 550 Max number of shared server sessions

object_cache_max_size_

percent10

Percentage of maximum size over optimal of the

user sessions obobject_cache_optimal_size

102400Optimal size of the user sessions object cache in

bytes

olap_page_pool_size 33554432 Size of the olap page pool in bytes

open_cursors 500 Max # cursors per session

open_links 4 Max # open links per session

open_links_per_instance 4 Max # open links per instance

optimizer_dynamic_

sampling1 Optimizer dynamic sampling

optimizer_features_

enable9.2.0 Optimizer plan compatibility parameter

optimizer_index_caching 0 Optimizer percent index cachingoptimizer_index_cost_adj 100 Optimizer index cost adjustment

optimizer_max_

permutations2000

Optimizer maximum join permutations per query

block

optimizer_mode CHOOSE Optimizer mode

oracle_trace_collection_name

Oracle TRACE default collection name

oracle_trace_collection_path

%ORACLE_HOME%\OTRACE\ADMIN\CDF\

Oracle TRACE collection path

oracle_trace_collection_size

5242880 Oracle TRACE collection le max. size

oracle_trace_enable FALSE Oracle Trace enabled/disabledoracle_trace_facility_name

oracled Oracle TRACE default facility name

oracle_trace_facility_path%ORACLE_HOME%\OTRACE\ADMIN\FDF\

Oracle TRACE facility path

os_authent_prex OPS$ Prex for auto-logon accounts

os_roles FALSE Retrieve roles from the operating system

parallel_adaptive_multi_user

TRUEEnable adaptive setting of degree for multiple userstreams

parallel_automatic_tuning TRUEEnable intelligent defaults for parallel executionparameters

parallel_execution_message_size 4096 Message buffer size for parallel execution

parallel_instance_group Instance group to use for all parallel operations

parallel_max_servers 40 Maximum parallel query servers per instance

parallel_min_percent 0Minimum percent of threads required for parallelquery


27/40


parallel_min_servers 0 Minimum parallel query servers per instance

parallel_server FALSE If TRUE startup in parallel server mode

parallel_server_instances 1 Number of instances to use for sizing OPS SGAstructures

parallel_threads_per_cpu 2 Number of parallel execution threads per CPU

partition_view_enabled FALSE Enable/disable partitioned views

pga_aggregate_target 1078984704Target size for the aggregate PGA memory

consumed by the instanc

plsql_compiler_ags INTERPRETED PL/SQL compiler ags

plsql_native_c_compiler PL/SQL native C compiler

plsql_native_library_dir PL/SQL native library dir

plsql_native_library_subdir_count

0 PL/SQL native library number of subdirectories

plsql_native_linker PL/SQL native linker

plsql_native_make_le_

name PL/SQL native compilation make le

plsql_native_make_utility PL/SQL native compilation make utility

plsql_v2_compatibility FALSE PL/SQL version 2.x compatibility ag

pre_page_sga FALSE Pre-page SGA for process

processes 500 User processes

query_rewrite_enabled FALSEAllow rewrite of queries using materialized views if

enabled

query_rewrite_integrity ENFORCEDPerform rewrite using materialized views with

desired integrity

rdbms_server_dn RDBMSs Distinguished Name

read_only_open_delayed FALSEIf TRUE delay opening of read only les until rst

access

recovery_parallelism 0Number of server processes to use for parallelrecovery

remote_archive_enable true Remote archival enable setting

remote_dependencies_mode

TIMESTAMPRemote-procedure-call dependencies modeparameter

remote_listener Remote listener

remote_login_passwordle

EXCLUSIVE Password le usage parameter

remote_os_authent FALSE Allow non-secure remote clients to use auto-logonaccounts

remote_os_roles FALSE Allow non-secure remote clients to use os roles

replication_dependency_tracking

TRUETracking dependency for Replication parallelpropagation

resource_limit FALSE Master switch for resource limit


28/40



resource_manager_plan Resource mgr top plan

rollback_segments Undo segment list

row_locking always Row-locking

serial_reuse DISABLE Reuse the frame segments

serializable FALSE Serializable

service_names StressDM Service names supported by the instance

session_cached_cursors 0Number of cursors to save in the session cursorcache

session_max_open_les 10 Maximum number of open les allowed per session

sessions 555 User and system sessions

sga_max_size 1114854220 Max total SGA size

shadow_core_dump partial Core Size for Shadow Processes

shared_memory_address 0 SGA starting address (low order 32-bits on 64-bitplatforms)

shared_pool_reserved_size

5452595 Size in bytes of reserved area of shared pool

shared_pool_size 109051904 Size in bytes of shared pool

shared_server_sessions 550 Max number of shared server sessions

shared_servers 1 Number of shared servers to start up

sort_area_retained_size 0Size of in-memory sort work area retained between

fetch calls

sort_area_size 524288 Size of in-memory sort work area

sple

%ORACLE_HOME%\DATABASE\

SPFILE%ORACLE_SID%.ORA

Server parameter le

sql92_security FALSE Require select privilege for searched update/delete

sql_trace FALSE Enable SQL trace

sql_version NATIVESQL language version parameter for compatibilityissues

standby_archive_dest%ORACLE_HOME%\RDBMS

Standby database archivelog destination text string

standby_le_

managementMANUAL

If auto then les are created/dropped automatically

on standby

star_transformation_

enabled

FALSE Enable the use of star transformation

statistics_level TYPICAL Statistics level

tape_asynch_io TRUE Use asynch I/O requests for tape devices

thread 0 Redo thread to mount

timed_os_statistics 0 Internal os statistic gathering interval in seconds

timed_statistics TRUE Maintain internal timing statistics


29/40


trace_enabled FALSE Enable KST tracing

tracele_identier Trace le custom identier

transaction_auditing TRUE Transaction auditing records generated in the redolog

transactions 610 Max. number of concurrent active transactions

transactions_per_rollback_segment

5 Number of active transactions per rollback segment

undo_management AUTOInstance runs in SMU mode if TRUE, else in RBUmode

undo_retention 10800 Undo retention in seconds

undo_suppress_errors FALSE Suppress RBU errors in SMU mode

undo_tablespace UNDOTBS1 Use/switch undo tablespace

use_indirect_data_buffers FALSEEnable indirect data buffers (very large SGA on 32-

bit platforms

user_dump_destE:\oracle\admin\StressDM\udump

User process dump directory

utl_le_dir utl_le accessible directories list

workarea_size_policy AUTOPolicy used to size SQL working areas (MANUAL/

AUTO)


30/40


31/40


control_les

E:\ORACLE\PRODUCT\10.2.0\

ORADATA\STRESS10\CONTROL01.CTL, E:\ORACLE\PRODUCT\10.2.0\ORADATA\STRESS10\CONTROL02.CTL, E:\ORACLE\PRODUCT\10.2.0\ORADATA\STRESS10\CONTROL03.CTL

Control le names list

core_dump_dest


ADMIN\STRESS10\CDUMP

Core dump directory

cpu_count 4 Number of CPUs for this instance

create_bitmap_area_size 8388608 Size of create bitmap buffer for bitmap index

create_stored_outlines Create stored outlines for DML statements

cursor_sharing SIMILAR Cursor sharing mode

cursor_space_for_time FALSE Use more memory in order to get faster execution






db_block_buffers 0 Number of database blocks cached in memory

db_block_checking FALSE Header checking and data and index block checking

db_block_checksum TRUE Store checksum in db blocks and check during reads

db_block_size 8192 Size of database block in bytes

db_cache_advice ON Buffer cache sizing advisory

db_cache_size 0Size of DEFAULT buffer pool for standard block size

buffers

db_create_le_dest Default database location



db_create_online_log_dest_2 Online log/controlle destination #2






32/40





db_domain Directory part of global database name stored withCREATE DATABA

db_le_multiblock_read_

count8 DB block to be read each IO

db_le_name_convert Datale name convert patterns and strings for

standby/clone db

db_les 200 Max allowable # db les

db_ashback_retention_

target1440

Maximum Flashback Database log retention time inminutes.

db_keep_cache_size 0Size of KEEP buffer pool for standard block size

buffers

db_name Stress10 Database name specied in CREATE DATABASE

db_recovery_le_destE:\oracle\product\10.2.0/ash_recovery_area

Default database recovery le location

db_recovery_le_dest_

size2147483648 Database recovery les size limit

db_recycle_cache_size 0Size of RECYCLE buffer pool for standard block size

buffers

db_unique_name Stress10 Database Unique Name

db_writer_processes 2Number of background database writer processes tostart

dbwr_io_slaves 0 DBWR I/O slaves

ddl_wait_for_locks FALSE Disable NOWAIT DML lock acquisitions

dg_broker_cong_le1

E:\ORACLE\PRODUCT\10.2.0\DB_1\DATABASE\DR1STRESS10.DAT


dg_broker_cong_le2

E:\ORACLE\PRODUCT\10.2.0\DB_1\DATABASE\DR2STRESS10.DAT


dg_broker_start FALSE Start Data Guard broker framework (DMON process)

disk_asynch_io TRUE Use asynch I/O for random access devices

dispatchers(PROTOCOL=TCP)SERVICE=Stress10XDB)


distributed_lock_timeout 60Number of seconds a distributed transaction waitsfor a lock

dml_locks 748DML locks - one for each table modied in a

transaction

drs_start FALSE Start DG Broker monitor (DMON process)

event Debug event control - default null string


33/40

9


fal_client FAL client

fal_server FAL server list

fast_start_io_target 0 Upper bound on recovery reads

fast_start_mttr_target 0 MTTR target of forward crash recovery in seconds

fast_start_parallel_rollback

LOWMax number of parallel recovery slaves that may beused

le_mapping FALSE Enable le mapping

leio_network_adapters Network Adapters for File I/O

lesystemio_options IO operations on lesystem les

xed_date Fixed SYSDATE value

gc_les_to_locks Mapping between le numbers and global cache

locks

gcs_server_processes 0 Number of background gcs server processes to start

global_context_pool_size Global Application Context Pool Size in Bytes

global_names FALSEEnforce that database links have same name asremote database

hash_area_size 131072 Size of in-memory hash work area

hi_shared_memory_address

0SGA starting address (high order 32-bits on 64-bitplatforms)

hs_autoregister TRUEEnable automatic server DD updates in HS agentself-registration

ile Include le in init.ora

instance_groups List of instance group names

instance_name stress10 Instance name supported by the instance

instance_number 0 Instance number

instance_type RDBMS Type of instance to be executed

java_max_sessionspace_size

0 Max allowed size in bytes of a Java sessionspace

java_pool_size 0 Size in bytes of java pool

java_soft_sessionspace_limit

0Warning limit on size in bytes of a Java

sessionspace

job_queue_processes 10 Number of job queue slave processes

large_pool_size 0 Size in bytes of large pool

ldap_directory_access NONE RDBMSs LDAP access option

license_max_sessions 0 Maximum number of non-system user sessionsallowed

license_max_users 0Maximum number of named users that can becreated in the databas

license_sessions_warning

0Warning level for number of non-system usersessions


34/40


0


local_listener Local listener

lock_name_spaceLock name space used for generating lock names

for standby/clonelock_sga FALSE Lock entire SGA in physical memory

log_archive_cong Log archive cong parameter

log_archive_dest Archival destination text string







log_archive_dest_6 Archival destination #6 text stringlog_archive_dest_7 Archival destination #7 text string




log_archive_dest_state_10

enable Archival destination #10 state text string









log_archive_duplex_dest Duplex archival destination text string

log_archive_format ARC%S_%R.%T Archival destination format

log_archive_local_rst TRUE Establish EXPEDITE attribute default value

log_archive_max_processes

2 Maximum number of active ARCH processes

log_archive_min_succeed_dest

1Minimum number of archive destinations that mustsucceed

log_archive_start FALSE Start archival process on SGA initialization

log_archive_trace 0 Establish archivelog operation tracing level

log_buffer 7024640 Redo circular buffer size

log_checkpoint_interval 0 # redo blocks checkpoint threshold


35/40


log_checkpoint_timeout 1800Maximum time interval between checkpoints inseconds

log_checkpoints_to_alert FALSE Log checkpoint begin/end to alert le

log_le_name_convert Logle name convert patterns and strings for

standby/clone db

logmnr_max_persistent_sessions

1 Maximum number of threads to mine

max_commit_propagation_delay

0 Max age of new snapshot in .01 seconds

max_dispatchers Max number of dispatchers

max_dump_le_size UNLIMITED Maximum size (blocks) of dump le

max_enabled_roles 150 Max number of roles a user can have enabled

max_shared_servers Max number of shared servers

object_cache_max_size_percent

10 Percentage of maximum size over optimal of theuser sessions ob

object_cache_optimal_size

102400Optimal size of the user sessions object cache in

bytes

olap_page_pool_size 0 Size of the olap page pool in bytes

open_cursors 300 Max # cursors per session

open_links 4 Max # open links per session

open_links_per_instance 4 Max # open links per instance

optimizer_dynamic_

sampling2 Optimizer dynamic sampling

optimizer_features_

enable

10.2.0.1 Optimizer plan compatibility parameter

optimizer_index_caching 0 Optimizer percent index caching

optimizer_index_cost_adj 100 Optimizer index cost adjustment

optimizer_mode ALL_ROWS Optimizer mode

optimizer_secure_view_

mergingTRUE

Optimizer secure view merging and predicate

pushdown/movearound

os_authent_prex OPS$ Prex for auto-logon accounts

os_roles FALSE Retrieve roles from the operating system

parallel_adaptive_multi_user

TRUEEnable adaptive setting of degree for multiple userstreams

parallel_automatic_tuning FALSEEnable intelligent defaults for parallel execution

parametersparallel_execution_message_size

2148 Message buffer size for parallel execution

parallel_instance_group Instance group to use for all parallel operations

parallel_max_servers 80 Maximum parallel query servers per instance


36/40



parallel_min_percent 0Minimum percent of threads required for parallelquery

parallel_min_servers 0 Minimum parallel query servers per instanceparallel_server FALSE If TRUE startup in parallel server mode

parallel_server_instances 1Number of instances to use for sizing OPS SGA

structures

parallel_threads_per_cpu 2 Number of parallel execution threads per CPU

pga_aggregate_target 419430400Target size for the aggregate PGA memory

consumed by the instanc

plsql_ccags PL/SQL ccags

plsql_code_type INTERPRETED PL/SQL code-type

plsql_compiler_agsINTERPRETED, NON_DEBUG

PL/SQL compiler ags

plsql_debug FALSE PL/SQL debugplsql_native_library_dir PL/SQL native library dir

plsql_native_library_subdir_count

0 PL/SQL native library number of subdirectories

plsql_optimize_level 2 PL/SQL optimize level

plsql_v2_compatibility FALSE PL/SQL version 2.x compatibility ag

plsql_warnings DISABLE:ALL PL/SQL compiler warnings settings

pre_page_sga FALSE Pre-page sga for process

processes 150 User processes

query_rewrite_enabled TRUEAllow rewrite of queries using materialized views if

enabled

query_rewrite_integrity enforcedPerform rewrite using materialized views with

desired integrity

rdbms_server_dn RDBMSs Distinguished Name

read_only_open_delayed FALSEIf TRUE delay opening of read only les until rst

access

recovery_parallelism 0Number of server processes to use for parallelrecovery

recyclebin off Recyclebin processing

remote_archive_enable true Remote archival enable setting

remote_dependencies_mode

TIMESTAMPRemote-procedure-call dependencies modeparameter

remote_listener Remote listener remote_login_passwordle

EXCLUSIVE Password le usage parameter

remote_os_authent FALSEAllow non-secure remote clients to use auto-logonaccounts

remote_os_roles FALSE Allow non-secure remote clients to use os roles


37/40


replication_dependency_tracking

TRUETracking dependency for Replication parallelpropagation

resource_limit FALSE Master switch for resource limitresource_manager_plan Resource mgr top plan

resumable_timeout 0 Set resumable_timeout

rollback_segments Undo segment list

serial_reuse disable Reuse the frame segments

service_names Stress10 Service names supported by the instance

session_cached_cursors 20 Number of cursors to cache in a session.

session_max_open_les 10 Maximum number of open les allowed per session

sessions 170 User and system sessions

sga_max_size 1660944384 Max total SGA size

sga_target 1048576000 Target size of SGAshadow_core_dump partial Core Size for Shadow Processes

shared_memory_address 0SGA starting address (low order 32-bits on 64-bitplatforms)

shared_pool_reserved_size

5452595 Size (in bytes) of reserved area of shared pool

shared_pool_size 0 Size in bytes of shared pool

shared_server_sessions Max number of shared server sessions

shared_servers 1 Number of shared servers to start up

skip_unusable_indexes TRUE Skip unusable indexes if set to TRUE

smtp_out_server utl_smtp server and port conguration parameter

sort_area_retained_size 0 Size of in-memory sort work area retained betweenfetch calls

sort_area_size 65536 Size of in-memory sort work area

sple

E:\ORACLE\PRODUCT\10.2.0\DB_1\DBS\SPFILESTRESS10.ORA

Server parameter le

sql92_security FALSE Require select privilege for searched update/delete

sql_trace FALSE Enable SQL trace

sql_version NATIVESQL language version parameter for compatibilityissues

sqltune_category DEFAULT Category qualier for applying hintsets

standby_archive_dest%ORACLE_HOME%\RDBMS

Standby database archivelog destination text string

standby_le_

managementMANUAL

If auto then les are created/dropped automatically

on standby

star_transformation_enabled

FALSE Enable the use of star transformation


38/40



statistics_level TYPICAL Statistics level

streams_pool_size 0 Size in bytes of the streams pool

tape_asynch_io TRUE Use asynch I/O requests for tape devices

thread 0 Redo thread to mount

timed_os_statistics 0 Internal os statistic gathering interval in seconds

timed_statistics TRUE Maintain internal timing statistics

trace_enabled FALSE Enable KST tracing

tracele_identier Trace le custom identier

transactions 187 Max. number of concurrent active transactions

transactions_per_rollback_segment

5 Number of active transactions per rollback segment

undo_management AUTOInstance runs in SMU mode if TRUE, else in RBUmode

undo_retention 900 Undo retention in seconds

undo_tablespace UNDOTBS1 Use/switch undo tablespace

use_indirect_data_buffers

FALSEEnable indirect data buffers (very large SGA on 32-bit platforms

user_dump_dest


ADMIN\STRESS10\UDUMP

User process dump directory

utl_le_dir utl_le accessible directories list

workarea_size_policy AUTOPolicy used to size SQL working areas (MANUAL/

AUTO)


39/40


40/40


Fdm 92000 Architecture Analysis Guide

Documents