Title of Your Presentation

Download Report

Transcript Title of Your Presentation

Machine Sizing and Scalability
Mark Nesson, Vashti Ragoonath
June 2008
Copyright 2007, Information Builders. Slide 1
Sizing and Scalability
Overview
The question of sizing and scalability always comes up.
 Let’s talk about sizing
 Why is it important
 What components need sizing considerations
 Machines
 Network bandwidth and traffic
 File Disks
 CPU’s and Speed
 WebFOCUS Repositories
 Resource Analyzer and Governor
 ReportCaster Repository
 Schedules
 Report Library Content
 Data Migrator
 MR Realm Driver Repository
Copyright 2007, Information Builders. Slide 2
Sizing and Scalability
Overview
 Let’s talk about Scalability
 What is scalability and why is it important
 Ability of a site/application to maintain performance,
reliability and availability as usage increases.
 Why does WebFOCUS scale so well
 Non-persistence
 Multi-Threading
 Native Data Adapters
 Superior Data Manipulation Technology
 Multi-Tier Configurations
 Clustered Servers and Load Management
Copyright 2007, Information Builders. Slide 3
Sizing and Scalability
Sizing Factors
One of the most common questions asked:
 How many machines and resources do I need?
 The WebFOCUS Architecture will determine how many
machines you need to begin with
 Benchmarks will determine the capacity of the machines to
achieve a scalable and HA system
 WebFOCUS Architecture factors to consider:
 Placement of Components
 Network bandwidth
 Failover and Load Balancing requirements
 Benchmark factors to consider:
 Database Capacity
 Queries
 Performance Criteria and Target Response Times
Copyright 2007, Information Builders. Slide 4
Sizing and Scalability
Machine Sizing Factors - Architecture
 WebFOCUS Architecture
 Central or distributed WebFOCUS configuration
 Central – all components on the same machine
 Distributed – components are installed on separate
machines
 Central Configuration
 Network traffic not an issue
 Contention for resources may be a problem
 Single point of failure unless Failover is implemented
 Benchmarks a determining factor in scalability
Copyright 2007, Information Builders. Slide 5
Sizing and Scalability
Machine Sizing Factors - Architecture

WebFOCUS Architecture
 Distributed Configuration
 Components should be installed with the network
bandwidth and traffic in mind
 Consider the following:
 Do you want any of the WF software on the same machine
as your Database (Oracle for example) ?
 Do you want the Application server on the same machine
as your Reporting Server?
 If ReportCaster is in the plan where would you like the
Distribution Server – it’s a standalone server so it can be
anywhere. Where is best?
 Is Failover to be implemented?
 Is Load Balancing to be implemented?
Copyright 2007, Information Builders. Slide 6
Sizing and Scalability
Capacity Sizing Factors - Database

Database Capacity
 What’s the network like between:
 The Reporting Server and the Databases
 The ReportCaster Distribution Server and the
ReportCaster DBMS repository
 The Application Server and the ReportCaster DBMS
repository
 The Application Server and the MR Realm Driver DBMS
repository
 Are the databases optimally tuned ?
 Are the databases indexed?
 Is the Database server configured to accept a limited number
of concurrent connections?
Copyright 2007, Information Builders. Slide 7
Sizing and Scalability
Capacity Sizing Factors - Queries


Queries and Performance Criteria
 What are the queries like
 How much data is retrieved
 Are the queries CPU intensive
 What is the expected average response time for a report
 Are they optimized
 Benchmark a fair representation of production queries to
determine scalability
 Repetitive process of tuning and benchmarking
We will be demoing the tuning and repetitive benchmarks in the
“Performance and Tuning” presentation
Copyright 2007, Information Builders. Slide 8
Sizing and Scalability
Repository Sizing

Now let’s move on to repository sizing.
 WebFOCUS has a number of repositories:
 Resource Analyzer and Governor
 ReportCaster Repository
 Schedule Information
 Report Library Content
 Data Migrator (ETL)
 MR Realm Driver Repository

We will focus on:
 Resource Analyzer
 ReportCaster
Copyright 2007, Information Builders. Slide 9
Sizing and Scalability
Repository Sizing –Resource Analyzer
To allocate adequate storage for RA, customer must know:
 What data is be captured and stored
 How will requests be launched
 How many requests will be run
 What number of COMPUTEs and DEFINEs are in the procedures
to be monitored
 What is the retention time for the repository data
 How often can the repository be purged
IBI can provide customer with guidelines:
 The row size for each of the repository tables
 Number of rows per table inserted per request
 Tables populated for each Monitor Preference
Copyright 2007, Information Builders. Slide 10
Sizing and Scalability
Demo: RA Storage Utilization


Let’s see a live demo of how much space we need for Resource
Analyzer
Our demo is based on a Monitor Preference of Query and Froms
because we wish to capture the query details, tables being
accessed and the user information
 Sample WebFOCUS request that contains one of each
statistic collected. The request contains:
 Number of:
 DEFINEs (1) .. counted as a COLUMN
 COMPUTEs(1) .. counted as a COLUMN
 RELATIONs(1)
 SQL Passthru (1)
 Real COLUMNS (8)
Copyright 2007, Information Builders. Slide 11
Sizing and Scalability
Demo: RA Storage Utilization

Let’s start…
 Capture the space utilized in the Oracle tablespace
 Run our sample query
 Capture the space utilized in the Oracle tablespace
 Run the Oracle ANALYZE command to see how now many
rows were populated for each table and number of bytes per
row
 Now we can use these numbers to get an approximation of
space to allocate for RA utilization (number of rows per table
by number of requests)
Copyright 2007, Information Builders. Slide 12
Sizing and Scalability
Repository Sizing –Report Library
To allocate adequate storage for RL, customer must know:
 Number of schedules with distribution to Report Library
 Distributed formats to the Report Library
 Scheduling from applet or HTML Wizard
 Archiving rules:
 How many versions of a report should be kept
 How long should versions be kept for
 How often can the repository be completely purged
Copyright 2007, Information Builders. Slide 13
Sizing and Scalability
Demo: RL Storage Utilization


Let’s start with uncompressed data…
 Capture the space utilized in the Oracle tablespace
 Schedule a job from Applet and distribute to the Report
Library in PDF format. Compression flag unchecked
 Capture the space utilized in the Oracle tablespace
 Run the Oracle ANALYZE command to see how now many
rows were populated for each table and number of bytes per
row
 We look specifically at the to check for size of file in the
BOTLDATA table and verify that it is flagged as
uncompressed data
Now we can use these numbers to get an approximation of space
to allocate uncompressed data for our example
Copyright 2007, Information Builders. Slide 14
Sizing and Scalability
Demo: RL Storage Utilization


Let’s start with compressed data..
 Capture the space utilized in the Oracle tablespace
 Schedule same job from Applet and distribute to the Report
Library in PDF format. Compression flag should be checked
 Capture the space utilized in the Oracle tablespace.
 Run the Oracle ANALYZE command to see how now many
rows were populated for each table and number of bytes per
row
 We look specifically at the to check for size of file in the
BOTLDATA table and verify that it is flagged as compressed
data
Now we can use these numbers to get an approximation of space
to allocate compressed data for our example
Copyright 2007, Information Builders. Slide 15
Sizing and Scalability
Conclusion


This presentation provides guidelines that will vary for each
customer and we hope it was useful
Questions and comments
Copyright 2007, Information Builders. Slide 16