You are on page 1of 16

Oracle Hyperion Financial Data Quality

Management, Enterprise Edition


Sizing Guide Performance Report
ORACLE WHITE PAPER | JUNE 2015

Disclaimer
This document is provided for information purposes and should not be relied upon in making a
purchasing decision. The contents hereof are subject to change without notice. This document is not
warranted to be error-free, nor subject to any other warranties or conditions, whether expressed orally
or implied in law, including implied warranties and conditions of merchantability or fitness for a
particular purpose.

THIS DOCUMENT IS NOT PART OF A LICENSE AGREEMENT NOR CAN IT BE


INCORPORATED INTO ANY CONTRACTUAL AGREEMENT WITH ORACLE
CORPORATION OR ITS SUBSIDIARIES OR AFFILIATES.
Failure to adhere to these benchmarks does not constitute a breach of Oracles obligations. We
specifically disclaim any liability with respect to this document and no contractual obligations are
formed either directly or indirectly by this document. This document may not be reproduced or
transmitted in any form or by any means, electronic or mechanical, for any purpose, without our prior
written permission.

ORACLE HYPERION FINANCIAL DATA QUALITY MANAGEMENT, ENTERPRISE EDITION - SIZING GUIDE

Table of Contents
Disclaimer

Introduction

User Load, Monitoring and Tuning Environment Variables

User Load

Monitor

Tuning Environment Variables

Test Hardware Configuration Specifications


Configuration Tuning

5
5

Tuned Settings for Test Configuration

Default Settings

Test scenarios

FDMEE Test Data Details

Results

Single User Test Results


Multi-User Test Results

8
11

Clustering Load Balancing

12

Database Sizing

12

Conclusion

12

Reference documents

13

1 | ORACLE HYPERION FINANCIAL DATA QUALITY MANAGEMENT, ENTERPRISE EDITION - SIZING GUIDE

Introduction
Finance organizations continually enhance the quality of their internal controls and reporting
processes. To meet these goals, a source-to-report view of financial data processes is required.
Oracle Hyperion Financial Data Quality Management, Enterprise Edition (FDMEE) is a solution that
allows business analysts to develop standardized financial data management processes and
validate data from any source systemall while reducing costs and complexity. FDMEE puts the
finance user in total control of the integration process to define source data, create mapping rules to
translate data into the required target format, and to execute and manage the periodic data loading
process.
This document provides some general guidance for sizing an FDMEE environment. It includes
guidance on resource usage, (CPU utilization, Heap memory), expected response times and where
to get more information on monitoring and tuning your system. A properly sized and tuned
environment ensures a consistent experience for all your systems users. Tuning and optimization of
other applications and databases is not covered under the scope of this paper.
Consider three factors when analyzing the capacity of your current environment to ensure it meets
your business requirements. Assessing both peak and off peak user load, monitoring the
environment and tuning environment variables after monitoring.
User Load - Evaluating peak and off-peak user loads can help to create a properly sized
environment.
Monitor - Monitor the environment to determine if adjustments are necessary.
Tuning environment Variables - Default values for the environment variables may need to be
adjusted for your specific user load.

Estimating for user load, monitoring the system and adjusting tuning variables are the keys to a
successful implementation. This document offers guidance on all three factors, and provides
examples of what you can expect. It helps you plan for your initial installation and suggests when to
make necessary adjustments to the resources supporting your installation for the future. Due to
variations in system usage and volumes of data, your results may be different than what is described
here.

Users of this guide should be familiar with Oracle Hyperion Financial Data Quality Management
Enterprise Edition, database administration, and general operating system concepts.

Always make sure the latest patches have been applied to all the servers in your configuration.

2 | ORACLE HYPERION FINANCIAL DATA QUALITY MANAGEMENT, ENTERPRISE EDITION SIZING GUIDE

User Load, Monitoring and Tuning Environment Variables


User Load
User load refers to the maximum number of users executing activities on the system at any given time.
Concurrency refers to the number of users that are executing activities simultaneously. User activities can be broken
down into two main categories; database read activities, for example running reports, and database write activities,
for example Loading Data.

TABLE 1 USER LOAD PLANNING


Total overall system users ______________

User Activity

% Concurrency

% Users during peak

Read (examples: reports, drillthrough)

(example: 25%)

(example: 50%)

Write (examples: create rules,


import mappings, load data)

(example: 25%)

(example: 50%)

Use this table to help with user load planning. % Concurrency ratio is the quantity of Read/Write users to the total number of
Read/Write users executing an activity simultaneously. % Users during peak ratio is the quantity of Read/Write users to the total
number of Read/Write users executing activities during peak times of the year.

Your goal should be to have your configuration operating about 75-85% capacity. The reason is that an overloaded
system does not operate efficiently. If your environment operates consistently at above 85% CPU or memory
capacity, consider whether your hardware or virtual images can support peak loads and future growth. Some
questions to consider when sizing a system:
1. How long does my system operate at peak load, One week per month? One week per quarter?
2. Are there periods where there is significantly more write versus read activities?

Monitor
Monitoring the environment during peak and off peak periods will provide you a greater understanding of your
current usage capacity, and helps you anticipate and plan for future capacity requirements. Several tools are
available to assist you in this activity.

Oracle Enterprise Manager Fusion Middleware Control


This tool allows you to manage, monitor and diagnose graphically the condition and status of your environment, in real time, while
the system is up and running. Enterprise Manager contains live statistics some of which are: % CPU utilization, heap memory and
datasources. For more information on using Oracle Enterprise Manager Fusion Middleware Control see:
http://docs.oracle.com/cd/E11857_01/index.htm

Windows Performance Monitor


The recommended tool to monitor Windows performance is Microsoft Windows Performance Monitor. This tool allows you to set
up a collection of data over time. You can collect many Windows metrics, among the most useful are % CPU utilization and Private
Bytes.

Linux Performance Monitor


For Linux there are many options for gathering system statistics, some of which include dstat, vmstat and iostat.

3 | ORACLE HYPERION FINANCIAL DATA QUALITY MANAGEMENT, ENTERPRISE EDITION SIZING GUIDE

Tuning Environment Variables


Many variables determine the performance of an FDMEE configuration. They can be monitored easily and to
correct many performance problems. The variables include:
CPU utilization (requires physical hardware upgrade)
Available RAM (requires physical hardware upgrade)

Heap size
Datasources
Database tuning
Number of concurrent users
Frequency of specific user activities
Import format, Data Load Mapping and script complexity

Database read activities are less resource intensive, if you find your Data Management process includes more read
activities then monitor CPU, heap and datasources. Then make adjustments to the following.
Heap size
WebLogic Datasources

Database write activities are more resource intensive; if you find your Data Management process includes more
write activities, then monitor the above statistics as well as database statistics. Then assess whether any of the
following parameters requires adjustment.
Heap size
WebLogic Datasources
FDMEE - Batch size
ODI
Array Fetch Size
Batch Update Size
Number of Sessions
Oracle Database
Sessions
Processes

For instructions on how to change the above parameters in the Oracle Hyperion Financial Data Quality
Management, Enterprise Edition (FDMEE) Tuning Guidelines document see:

http://www.oracle.com/technetwork/middleware/bi-foundation/fdmee-tuning-1112x-2349440.pdf

4 | ORACLE HYPERION FINANCIAL DATA QUALITY MANAGEMENT, ENTERPRISE EDITION SIZING GUIDE

Test Hardware Configuration Specifications


The test hardware configuration consisted of four physical servers with Windows Server 2008 R2 Enterprise Edition,
hosting a single FDMEE server and ODI agent. All machines have 64-bit processors. Hyperthreading was enabled
on all the machines doubling the number of virtual CPUs (vCPUs) as physical cores.
TABLE 2 TEST HARDWARE SPECIFICATIONS

Server

Processor

Memory

Function

Foundation Services

Intel Xeon X5670

72 GB*

Workspace, Foundation Services, Weblogic


Admin Server

144 GB*

FDMEE Web Server, ODI

144 GB*

HFM, Planning, Essbase Server

256 GB*

Oracle 11g R2 RDBMS

12 cores (24 vCPUs) @2.93 GHz


FDMEE Web Server

Intel Xeon X5675


12 cores (24 vCPUs) @3.07 GHz

HFM, Planning, Essbase


Server

Intel Xeon X5675

DB Server

Intel Xeon E5-2690

12 cores (24 vCPUs) @3.07 Ghz

16 cores (32 vCPUs) @2.9 GHz

* While these servers have large memory and CPU capacities, the results of the tests show only a small portion of this was required
to run successfully.

Configuration Tuning
EPM version 11.1.2.4 was used for testing. In some cases it may be necessary to alter the default settings for your
environment, to meet the needs based on user volume and hardware specifications. The following changes were
made to the tuning settings in the test configuration:
Tuned Settings for Test Configuration
FDMEE Server
Batch Size - 30000
Batch Timeout in Minutes - 30
ODI Server
OracleDIAgent - Maximum number of sessions - 1000
Data Server Array Fetch Size 30000
Data Server Batch Update Size - 30000
WebLogic
Aif_datasources - 1500
Odimasterrepository - 500
Registry Heap
-Xms= 6144m, -Xmx = 6144m
Oracle DB
Processes - 2000
Sessions 3040
Open Cursors - 5000

5 | ORACLE HYPERION FINANCIAL DATA QUALITY MANAGEMENT, ENTERPRISE EDITION SIZING GUIDE

Default Settings
FDMEE Server
Batch Size - 1000
Batch Timeout in Minutes - 30
ODI Server
OracleDIAgent Maximum number of sessions - 1000
Data Server Array Fetch Size 30
Data Server Batch Update Size - 30
WebLogic
Aif_datasources - 100
Odimasterrepository - 50
Registry Heap
-Xms= 4096, -Xmx= 4096

Test scenarios
Single User Volume File Load:
Various sample file sizes were loaded from a single user, using the Data Load Rule option from the Workflow
Tasks Menu: 6KB, 20KB, 50KB, 100KB, 500KB, 1,000KB, 2,000KB
User Trial Balance Report for current location with rules, Create PDF, Offline generating 748 pages.
Loading 1, 3, 6 and 12 months of data from Oracle EBS Ledger.
Single user tests were manually executed from a Windows 7 Professional client machine with Microsoft Internet
Explorer 9.

Multi-User User File Load:


Various user loads 5, 10, 20 and 50 users importing a 6KB file from the Data Load Rule option on the Workflow
Tasks Menu.
Client loads for the multi-user tests were simulated using Oracles load testing tool, Oracle Application Testing Suite
(OATS). Two components of OATS were employed. Open Script allows users to record user actions (HTTP
requests) to simulate virtual users accessing the application simultaneously. It also allows for parameter
substitutions. The Oracle Load Test component creates a simulation of multiple users, or a scenario, where you can
manipulate a variety of parameters such as think time and user ramp up rate.
Oracle Application Testing Suite is an effective way to baseline the performance of an initial configuration, and then
using that baseline as a comparison as the configuration/system matures.

For more information on Oracle Application Testing Suite, see: http://www.oracle.com/technetwork/oem/apptest/etest-101273.html

Batch load and Data Load Workbench import tests were not performed during the multi-user testing. If users
perform these activities the resource usage pattern will look different than the results presented here.

6 | ORACLE HYPERION FINANCIAL DATA QUALITY MANAGEMENT, ENTERPRISE EDITION SIZING GUIDE

FDMEE Test Data Details


Location: File type is Fixed, Target application is Oracle Hyperion Financial Management (HFM).
Import Format: 14 mapping rows.
Data Load Mapping for Location: 211 rows of account mappings using LIKE, BETWEEN and EXPLICIT.
140 LIKE
13 BETWEEN
58 EXPLICIT
Note about Data Load Mappings No Multi-Dimension mappings were used in the test system. The number of
mappings and the type of mappings can greatly affect your performance. The order of mapping resource intensity
from lowest to highest is: Like, Between, Multi-Dimension, In and Explicit.
The target Oracle Hyperion Financial Management (HFM) application contains: 1,352 Accounts, 1,219 Entities

TABLE 3 VOLUME FILE LOAD, ROW COUNT

File Rows

TDATASEG Row Count


After Load

File Size

6K

3,278

540 KB

20K

10,963

1,823 KB

50K

27,065

4,496 KB

100K

70,789

9,522 KB

500K

353,332

47,513 KB

1M

707,032

95,076 KB

2M

1,414,064

190,151 KB

This table contains the details of the files used to run the single user, for varying volume sizes. It also contains the row count from
the TDATASEG table after the load is complete. Input files used were fixed width Trial Balance Reports.

The Oracle E-Business Suite (EBS) Financials source system is located in a data center separate from the test
configuration.

7 | ORACLE HYPERION FINANCIAL DATA QUALITY MANAGEMENT, ENTERPRISE EDITION SIZING GUIDE

Results
Results of the tests described above include some sample response times and metrics for CPU utilization, memory
and datasource connections.

Single User Test Results

TABLE 4 TEST CONFIGURATION - SAMPLE RESOURCE USAGE SINGLE USER

User Activity

aif_datasource
connections used

odiMasterRepository
connections used

Heap Memory
used - MB

Load a 20k file, 1 period, replace method,


Data Load Rule

500-700

Load a 500k file, 1 period, replace method,


Data Load Rule

600-800

Load 20k file, 1 period, online, replace


method, Data Load Workbench, import

500-600

Run Base Trial Balance report, producing


748 pages in PDF format, online

300-400

Run Base Trial Balance report, producing


748 pages in PDF format, offline

1,000-2,000

This table shows a datasource and heap memory usage for five sample user activities.

CPU Utilization for a single user activity will have a CPU spike, which in the test environment ranged from 40-50% of
a single CPU. The spike on the FDMEE server ranges from 5-15 seconds in duration after which the CPU activity
starts to rise on the database server. In the test environment, the CPU spike on the database server typically lasted
for 10-15 seconds. The size and duration of the spike varies with the amount of data you are loading.
Datasource utilization varies based on the user activity that is executing. You should note that any activities where
the method selected is online require more datasource connections as noted above in the table.

TABLE 5 TEST CONFIGURATION - SAMPLE RESPONSE TIMES SINGLE USER

User Activity

Average
Response Time, Sec.*

Load 20k file for one period, offline, using replace


method through Data Load Workbench, import only

15

Run Base Trial Balance report, producing 105 pages


in PDF format, online/offline

Run Base Trial Balance report, producing 748 pages


in PDF format, online/offline

25

* Average response time is calculated by taking the average of 5 response times for this activity.

8 | ORACLE HYPERION FINANCIAL DATA QUALITY MANAGEMENT, ENTERPRISE EDITION SIZING GUIDE

FIGURE 1 TEST CONFIGURATION - SAMPLE RESPONSE TIMES SINGLE USER VARYING FILE SIZE

Response Time - Single User


Import from File - Export to HFM
0:00
24:00
21:00

20:34

M 18:00
M 15:00
:
12:00
S
9:00
S

10:44
5:33

6:00
3:00

0:32

0:41

1:15

20k

50k

100k
500k
File Row Size

0:00

1M

2M

Data Load Rule - Load file, 1 period, Replace, Import from source, Export to target
This figure represents a single-user executing File Import through the Data Load rule.

Response times remain under 2 minutes for a file size of 100k rows or less. Factors that contribute to these
response times are available heap memory set in the Window registry and Batch Size as set in FDMEE System
Settings.
FIGURE 2 TEST CONFIGURATION - SAMPLE RESPONSE TIMES SINGLE USER DATA LOAD FROM EBS

Response Time - SingleUser


Import data from Oracle EBS GL

M
M
:
S
S

0:24:00
0:22:00
0:20:00
0:18:00
0:16:00
0:14:00
0:12:00
0:10:00
0:08:00
0:06:00
0:04:00
0:02:00
0:00:00

0:22:57
0:19:45

0:09:01
0:03:57

121,519

374,978

730,864

1,404,753

Rows loaded to TDATASEG table


The single-user test graph for data import from Oracle EBS GL shows sample times to load x number of rows. The bars correspond
to loading 1, 3, 6 and 12 months of data to the test system HFM application.

9 | ORACLE HYPERION FINANCIAL DATA QUALITY MANAGEMENT, ENTERPRISE EDITION SIZING GUIDE

These are sample approximations only, due to the Ledger system location outside of the network segment where
the test EPM system was located. These response times depend heavily on tuning modifications made to the Array
Fetch Size and Batch Update Size set within ODI. Each bar corresponds to loading 1, 3, 6 and 12 months of data
from the EBS GL.

FIGURE 3 TEST CONFIGURATION - SAMPLE CPU UTILIZATION SINGLE USER DATA LOAD FROM EBS

HyS9aifWeb - % Processor Time


12 month data load from EBS
70
60
50
40
30
20
10
0

HyS9aifWeb - % Processor Time

This figure shows the sample CPU utilization for the web application on the FDMEE test server for data import from Oracle EBS GL
for 12 months. The first CPU peak was a single month load before the 12 month test was executed.

CPU utilization for the FDMEE server during a single user, 12- month data load (1,404,753 rows) from Oracle EBS
ranged between 20-40% of a single CPU. The number of CPU spikes correlate to the number of periods chosen to
load from the source system. Each spike in the test system lasted for approximately 5 seconds.

10 | ORACLE HYPERION FINANCIAL DATA QUALITY MANAGEMENT, ENTERPRISE EDITION SIZING GUIDE

Multi-User Test Results


TABLE 6 - TEST CONFIGURATION - SAMPLE RESOURCE USAGE 5, 10, 20, 50 USERS

Load a 6k file for one period, using replace method through Data Load Rule, 5 iterations each user

User Activity

5 users

10 users

20 users

50 users

aif_datasource connections used

4-7

3-7

7-10

10-15

odiMasterRepository connections
used

4-8

4-12

4-19

10-20

% CPU Utilization

20-75

40-175

50-200

200-400

Heap Memory used - MB

Reached a high of
4,586 MB

Reached a high of
5,257 MB

Reached a high of
4,730 MB

Reached a high of
5,500 MB

The multi-user test load table for File Import through the Data Load rule shows the resource utilization at varying levels of virtual
users. Heap memory usage may not be incremental based on the number of users, and is dependent on garbage collection
schemes.
Note- 100% CPU utilization = 1 CPU, 200% denotes 2 CPUs etc.

FIGURE 4 TEST CONFIGURATION - SAMPLE RESPONSE TIMES MULTI-USER

Response Time - Multi-User


Import 6K file, Export to HFM
0:03:43

0:04:00
0:03:30
M
M
:
S
S

0:03:00
0:02:30
0:02:00
0:01:14

0:01:30
0:01:00

0:00:29

0:00:30

0:00:28

0:00:30

0:00:00
MM:SS
Data Load Rule - Load 6K file, 1 period, Replace, Import from source, Export to target
1 user

5 users

10 users

20 users

50 users

The multi-user test graph for File Import through the Data Load rule shows the response times under 2 minutes up to 20 users.

11 | ORACLE HYPERION FINANCIAL DATA QUALITY MANAGEMENT, ENTERPRISE EDITION SIZING GUIDE

Clustering Load Balancing


Clustering a configuration is another way to scale and extend your configuration to improve performance. If your
server has become CPU bound, adding another server to the configuration and creating another FDMEE web app
may be the method of choice. You can use the same method if you are using virtual machines, and your existing
VM cannot be extended. Clustered configurations distribute workloads among multiple identical cluster member
instances. This effectively multiplies the amount of resources available to the distributed process, and provides for
seamless scalability and fail over for high availability. In order to achieve your performance goals for the
environment, creating multiple instances of the FDMEE web app and load balancing them may be all that is
required. For more information on Clustering and Load Balancing see:

Oracle Enterprise Performance Management System - Deployment Options Guide:


http://docs.oracle.com/cd/E57185_01/epm.1112/epm_deployment_options.pdf

Database Sizing
Determine database space requirements by the amount of data to pass through the FDMEE system. For example
in the test system, the Oracle database TDATASEG table contained 8.6 million rows and the TDATAMAPSEG
contained 386,000+ rows. This row count represents repeatedly importing data files with 1-2 million rows as well as
moving 12 months of data from an external ledger system to one target application. The Oracle Database
tablespace was allocated 32GB of space and 92% is currently used. The test system Oracle database is
configured with Automatic Memory Management and FDMEE and ODI were configured with a tablespace separate
from all other components in the configuration. The table size example shown here may vary greatly from your
system based on the mappings, source and target systems in your environment. Consult your DBA to tune your
database to meet your configuration requirements.

Conclusion
The primary objective for FDMEE is to integrate and validate data from varying sources and move the data to target
systems. Reading and writing large data sets requires sizing an environment properly to ensure all users of the
system have a uniform performance experience. Based on the test configuration and the tests outlined above the
test system comfortably sustains a user population of up to 50 users all executing modest size write and read
activities.
It is important to bear in mind that the data presented here is a sample from one configuration, whose environmental
infrastructure, network, hardware, etc., may vary widely from yours. It should be used as a reference point to
compare to your configuration experiences. The tests included files and source data that varied enough to cover
small and large application usage for FDMEE.
The best way to identify the specific requirements of your system is to perform a load test on the environment with
use cases that reflect actual user flow. Included in this document are additional references to documentation that
are available to assist you with your sizing task.

12 | ORACLE HYPERION FINANCIAL DATA QUALITY MANAGEMENT, ENTERPRISE EDITION SIZING GUIDE

Reference documents
Oracle Hyperion Financial Management (HFM) Performance Report
Oracle Support Document 1994672.1 (Hyperion Financial Management (HFM) 11.1.2.4 Performance Report) can
be found at: https://mosemp.us.oracle.com/epmos/faces/DocumentDisplay?id=1994672.1
Oracle Hyperion Financial Management, Fusion Edition Performance Tuning Guide
The tuning guide can be found at: http://www.oracleimg.com/technetwork/middleware/financialmanagement/overview/hfmperformancetuning-1934817.pdf
Oracle Profiler
Oracle Jrockit Mission Control provides profiling capabilities for processes using Jrockit JVM which can be found
at: http://www.oracle.com/technetwork/middleware/jrockit/overview/missioncontrol-whitepaper-june08-1130357.pdf?ssSourceSiteId=ocomen
Oracle Database Tuning
Consult your DBA for tuning. Oracle Database Online Documentation 11g Release 2 (11.2) which can be found
at: https://docs.oracle.com/cd/E11882_01/server.112/e41573/toc.htm
For other Oracle Documentation
Oracle Documentation Library (http://www.oracle.com/technology/documentation/epm.html) on Oracle
Technology Network for other performance tuning or configuration guides.

13 | ORACLE HYPERION FINANCIAL DATA QUALITY MANAGEMENT, ENTERPRISE EDITION SIZING GUIDE

Oracle Corporation, World Headquarters

Worldwide Inquiries

500 Oracle Parkway

Phone: +1.650.506.7000

Redwood Shores, CA 94065, USA

Fax: +1.650.506.7200

CONNECT W ITH US

blogs.oracle.com/oracle

twitter.com/oracle

Copyright 2014, Oracle and/or its affiliates. All rights reserved. This document is provided for information purposes only, and the
contents hereof are subject to change without notice. This document is not warranted to be error-free, nor subject to any other
warranties or conditions, whether expressed orally or implied in law, including implied warranties and conditions of merchantability or
fitness for a particular purpose. We specifically disclaim any liability with respect to this document, and no contractual obligations are
formed either directly or indirectly by this document. This document may not be reproduced or transmitted in any form or by any
means, electronic or mechanical, for any purpose, without our prior written permission.

oracle.com

Oracle and Java are registered trademarks of Oracle and/or its affiliates. Other names may be trademarks of their respective owners.

facebook.com/oracle

Intel and Intel Xeon are trademarks or registered trademarks of Intel Corporation. All SPARC trademarks are used under license and
are trademarks or registered trademarks of SPARC International, Inc. AMD, Opteron, the AMD logo, and the AMD Opteron logo are
trademarks or registered trademarks of Advanced Micro Devices. UNIX is a registered trademark of The Open Group. 0615

You might also like