Sei sulla pagina 1di 10

LOGI XML W HITE PAPER | A NEW MODEL FOR CORP ORATE SUCCESS

WHITE PAPER

What Determines Your Report Response Time and


Performance?
From hardware and network bandwidth to factors that a developer can
control, learn what can boost your report response time and performance,
while also contributing to a richer end-user experience.

Abstract

What determines a reporting system’s speed and performance? The first things that always come
to mind are hardware (memory and processor), bandwidth and number of users. While these
factors undoubtedly play a role in system performance, they are typically outside of the report
developer’s control. However, there are other factors that the developers may successfully
employ to boost the speed and the performance of the system. These are mostly pertinent to the
environments of Web-based BI, and they are doubly beneficial in the sense that they not only
make the system more efficient, but also provide the end-user with a much richer BI experience.

 2009, LOGIXML, INC. PAGE 1


LOGI XML W HITE PAPER | A NEW MODEL FOR CORP ORATE SUCCESS

Contents
Executive Summary ....................................................................................................... 3
Introduction.................................................................................................................. 4
“Environmental” Factors ................................................................................................. 4
Hardware (CPU/RAM) ................................................................................................. 4
Network Bandwidth .................................................................................................... 4
Operating System and Database Platform ..................................................................... 5
Number of Simultaneous Users.................................................................................... 5
Data Schema ............................................................................................................ 5
Data Volume ............................................................................................................. 6
Report Definition Complexity ....................................................................................... 6
Number of Reports..................................................................................................... 6
Type of Application .................................................................................................... 6
Report Styles ............................................................................................................ 7
The Factors in the Developer’s Control ............................................................................. 7
Interactive Paging...................................................................................................... 7
Dynamic Sorting ........................................................................................................ 7
Drill-down/Drill-through Reporting (On-demand Data).................................................... 7
Linked Data Layers .................................................................................................... 7
Reusing Elements (Shared Elements) ........................................................................... 8
Dashboard Visualization ............................................................................................. 8
Ajax Controls and Actions ........................................................................................... 8
Dynamic Filtering....................................................................................................... 8
Dynamic Grouping and Aggregations ............................................................................ 8
Crosstab Reports ....................................................................................................... 9
Hybrid Data Processing Engine .................................................................................... 9
Conclusion .................................................................................................................... 9
About LogiXML .............................................................................................................. 9
Contact Us Today! ................................................................................................... 10

 2009, LOGIXML, INC. PAGE 2


LOGI XML W HITE PAPER | A NEW MODEL FOR CORP ORATE SUCCESS

Executive Summary
What contributes to reporting response times and overall performance? Traditionally, we
are inclined to think that hardware (CPU/RAM) and network bandwidth are the key
factors that make or break reporting performance. While these elements bear a good
amount of weight, there are in reality a host of other factors that should be taken into
consideration. These range from OS and database platform to number of concurrent
users, from database schema (or source schema) to data vol umes, and from the number
of reports to type of reporting application.

However, it is important to note that there are elements that determine reporting
response time and performance that are actually in the control of the report developer .
This is all the more true within a Web application model: factors such as shared
elements, dynamic filtering and type of data-processing engine allow the developer’s
decisions to have an impact on report performance.

It is therefore possible to divide the factors contributing to report performance into two
broad categories: those that developers cannot control, and those that developers can.
Furthermore, there are some unique aspects of Web-based reporting that have a
substantial impact on fast and efficient report performance. These include interactive
paging, dynamic sorting, drill-down or drill-through on-demand data, Dashboarding and
more. Moreover, many of these features also contribute to a richer end -user experience.

In all, the choices that a report developer makes can have a substantial impact on report
performance, regardless of the factors that are out of his control.

 2009, LOGIXML, INC. PAGE 3


LOGI XML W HITE PAPER | A NEW MODEL FOR CORP ORATE SUCCESS

Introduction
Within a reporting and analysis architecture, it is often difficult to identify all the
variables that contribute to reporting speed and performance.

In this white paper, we’ll take a closer look at all these factors, which together
determine the response time and performance of reporting. These factors can be
grouped into two categories—namely, those outside of the developer’s control, which we
could call “environmental factors,” and those within it.

Also, we will identify some areas of Web-based reporting that present some unique
performance advantages, and point at the ways in which the report developer can use
them to maximize their impact.

“Environmental” Factors
Report speed or performance depends on several factors that have to do with the
hardware “environment” of the system executing the report. Although the report
developer does not have direct control over them, it is important to identify them as well
as some of the metrics associated with them.

In particular, this factors are hardware, (CPU/RAM), network bandwidth, OS and


database platform, number of concurrent users, data scheme, data volumes, report
definition and complexity, number of concurrent reports, type of application and report
styles. Let us explore each of these aspects separately.

H a r d wa r e ( C P U / R A M )

It is obvious that the more robust a hardware solution, the better the overall report
performance will be. More specifically, the factor that weighs the most on how hardware
configuration affects report speed is a central processing unit (CPU) or processing
power, as well as available memory (RAM).

The CPU is the core of a computing machine, and it is the ele ment responsible for
executing the sequences of instructions. RAM (random access memory) is the amount of
memory that the computing machine can store and that can be accessed “at random,”
meaning free from a particular order or sequence.

The faster the processor (or processors), the quicker and more efficient the reporting in
terms of concurrent users and page views. In summary, for data size and complexity,
memory and file system processing play a large role in determining the response time
and performance benchmarks of a report.

N e t wo r k B a n d wi d t h

Connectivity to the reporting server will play a large role in how responsive each report
request will be. The points of connectivity are several, and include the following:

 End-user connectivity

 Database server connectivity

 2009, LOGIXML, INC. PAGE 4


LOGI XML W HITE PAPER | A NEW MODEL FOR CORP ORATE SUCCESS

 Application server connectivity

 Data warehouse server connectivity

 Reporting Web server connectivity

In all cases, the higher the bandwidth capacity, the shorter the response times will be at
report running time.

O p e r a t i n g S y s t em a n d D a t a b a s e P l a t f o r m

The operating system and database system setups have an impact on reporting
performance. More specifically, the main factors in this regard are the overall file
management performance and data load performance, as well as connectivity to the
database (via OLEDB, ODBC or other means).

N um be r of Sim ult a n e ou s Us e rs

The more users a system has to support at the same time, the more load will be placed
on the report server—and the longer the response times. As with most Web-based
applications, user load generally shows a linear growth when it comes to response
times—in other words, for any additional concurrent user, the overall speed of the
system will slow down in a lineally-proportional way.

Here is a real-case example using easily-understood metrics, based on the execution (in
seconds) of a single 10,000-record report:

 10 users: 2.86 seconds

 50 users: 4.56 seconds

 100 users: 8.50 seconds

 250 users: 14.64 seconds

Clearly, the other variable in this case that could be increased or decreased is the
number of records in the report—which (as we will see under “data volume”) also affects
report response time.

D a t a S c h em a

Complexity of the database schema or overall data source schemas will contribute to
performance based on metrics for query responsiveness and data set responsiveness.
Factors that may contribute to better data schema performance include the following:

 Well-defined indexes and database relationships

 Highly relational models

 Data warehouse models optimized around desired output (i.e. report -centric
rather than data-centric models)

 2009, LOGIXML, INC. PAGE 5


LOGI XML W HITE PAPER | A NEW MODEL FOR CORP ORATE SUCCESS

D a t a V o lum e

Data volume, or the amount of data that burdens your processor, is measured in three
dimensions:

 Number of data columns

 Number of data rows

 Size/type of data value

For instance, a one-row data set with 10,000 columns will have the same impact on
performance as a 1-column data set with 10,000 rows.

In some cases, a data server can adjust automatically and dynamically the way in which
data is loaded and processed every time a new data set is requested. An e xample of
such server is LogiXML’s hybrid data engine, which combines in-memory processing with
XML file-system processing, as we will see in more detail under the heading “Hybrid Data
Processing Engine.”

R e p o r t D e f i n i t i o n C om p l e x i t y

What affects the report definition complexity is the number of elements and attributes in
the report. As with any Web page, the more information requested for presentation, the
longer the file will take to load.

In LogiXML’s Web-based architecture, the end result of any report request will either be
HTML or any of LogiXML’s report formats such as PDF, Excel, Word, XML, CSV, etc. The
more complex a report definition, the larger the output of the file will be.

A single table in a report may produce an HTML file that is 30KB in size; add 10 more
tables and charts and now the HTML file will be 200 KB in size. As more complex reports
are being designed, the size of the output will grow and directly impact the overall load
times for the end user.

N um be r of R ep o rt s

Similarly to number of concurrent users, the number of overall reports will also
contribute to the performance of any reporting system. With multiple reports, the
likelihood exists that these reports will be executed or run at some point on the system,
impacting the overall performance.

T yp e of Ap p l ic at i o n

Reporting applications serve many purposes, so the use case of the application will also
impact performance. Some contributing factors with application design: how many users,
how much data, how often are these reports generated or executed, whether reports
scheduled are static, archived or dynamic.

 2009, LOGIXML, INC. PAGE 6


LOGI XML W HITE PAPER | A NEW MODEL FOR CORP ORATE SUCCESS

R e p o rt St yl e s

Related to report complexity is the style of reports, which is another factor impacting
performance. However, style is a separate component, since reporting cont ent can draw
styles from both HTML design as well as images, CSS and style-sheets. These files can
be contained as part of the reporting application or can be used as more dynamic links
and http/URL reference within the context of a report request.

The Factors in the Developer’s Control


As we have seen, the factors listed above are not in the developer’s control, but are part
of the “environment” of the system where the reporting takes place. Next, we will
expand upon the elements within a Web application model on which the developer has
some control, and see how they too affect reporting speed and performance.

It is important to note that many of these elements are unique to the LogiXML pure
Web-based approach to report design and development.

I nt er a ct i ve P a g in g

All reports and data-sets to be presented to the end-user are prepared as a set of
“pages” that can be viewed and navigated through dynamically within the browser. This
allows the server to only load the proper amount of data during the initial di splay of a
report and use the interactive controls to jump from page to page, or skip directly to a
specific location. This feature lightens the load on the system and enables faster report
navigation.

D yn am ic S ort i n g

Dynamic sorting allows users to return complete data sets (or paged data-sets) and
dynamically sort or re-sort on any value within the current display.

D r i l l - d o wn / D r i l l - t h r o u g h R e p o r t i n g ( O n - d e m a n d D a t a )

This feature allows data to be displayed as layers using dynamic drill -down and drill-
through options.

Drill-down and drill-through capabilities offer many benefits. From the end-user’s
perspective, they let the report-consumer conveniently view different levels of the report
at the click of a mouse (with drill-down); or access related reports or analyses without
leaving the page (with drill-through).

From a technical side, a drill-down and drill-through feature makes the report faster and
more efficient: it lightens the reporting load by only showing the current state of
requested data to the user at any given time.

L i n k ed D at a L a yer s

This feature allows creation and sharing of data sets across a single report or multiple
reports. It makes reporting more efficient by reusing of data and preventing multiples

 2009, LOGIXML, INC. PAGE 7


LOGI XML W HITE PAPER | A NEW MODEL FOR CORP ORATE SUCCESS

calls to the database to present similar (or the same) data in different ways to end-
users.

R e u s i n g E l e m e n t s ( S h a r e d E l em e n t s )

Reusing or sharing elements within a report allows for more complex reporting
environments, custom templates and highly-dynamic ways to build new reporting
content during report design-time.

D a s h bo a rd V is u a li za t io n

Dashboards make it possible to present data in a fully dynamic, unified way that is
highly customizable and personalized for each user within the system —in spite of the
data coming from the same report(s). This means that while the information presented
to the end-users is varied and can be tailored to their specific needs, the actual
reporting burden is considerably lighter.

On one hand, with dashboards the end-user gets the information most relevant to his
business on a single Web page—and, with more sophisticated solutions, even the ability
to act upon the information presented. On the other, this feature lightens the load on
the system by displaying only the data called up by the user when the dashbo ard is
displayed in his Web-browser.

Furthermore, dashboards also serve to effectively present grouped data, aggregations,


KPIs, etc., and can be the mechanism for creating a drill -down or multi-layered approach
to reporting. This further compounds the technical benefits of this feature, making
dashboards a powerful way to both display data and lighten the load on the system.

Aj a x C o nt r o ls a n d Act io n s

Leveraging Ajax scripts and actionable items within a report allows for highly dynamic
script-based processing to occur on the Web server directly instead of relying on this
processing to occur on the client browser.

In other words, with Ajax scripts and actionable items, it is only the subset of the page
being viewed that is refreshed rather than the whole Web page. This contributes to a
faster and more efficient report performance.

D yn am ic F i lt er i n g

Creating highly customizable user-input and filter controls allows production of desired
outputs for various users and groups. This also enables security manag ement as well as
dynamically presenting data that can be changed by the end -users within the browser.

D yn am ic G r ou p i n g and A g gr e g at i on s

Grouping of data to produce dynamic aggregated values that present both flat and
hierarchical displays of reporting information also makes reporting more efficient.

 2009, LOGIXML, INC. PAGE 8


LOGI XML W HITE PAPER | A NEW MODEL FOR CORP ORATE SUCCESS

C r os st a b R e p ort s

Pivot and crosstab logic allows for multi-dimensional analysis, grouping and aggregation
of data to take large relational volumes of data and analyze according to your user
specifications.

H yb r i d D at a Pr o c ess i ng E n g i ne

The type of processing engine employed for reporting and analysis makes a substantial
difference in both report performance and scalability. To leverage the full power of a
server, it is necessary to go beyond in-memory data processing. For larger data-sets,
hybrid engines such as the Logi engine become highly useful: these combine in -memory
processing with XML file-system processing.

Depending on the report size and the number of users that need to be supported, a
hybrid-type engine will automatically provide the scalability needed by shifting from in -
memory to file-system processing. This is a fully configurable parameter by the
developer, who can determine how much data should be processed in memory and when
the switch to XML file-system should take place.

Data Integration

Intelligent “data marts” and data integration tools can allow you to optimize your operational
data sources for BI. With tools such as Logi ETL, you can bring together data from various
heterogeneous data sources and quickly design a highly optimized data mart for your
reporting and analysis needs. This will allow you to better manage the load on your
transactional databases and present pre-aggregated, pre-customized data back to the
reporting front-end. Through data integration, developers and system administrators can take
advantage of increased scalability and performance to serve more reports to more users at
faster speeds without affecting operational database

Conclusion
As we have seen, the factors that contribute to a reporting system’s performance can be
divided between those that are outside of the developer’s control and those that depend
on him. Within the realm of Web-based BI, many of the factors on which a developer can
act have a double benefit.

On one hand, they contribute to faster and more efficient reporting, mainly (but not
only) by displaying on the user’s Web browser only the data or information that the user
has called up. On the other, they make up for a richer, more interactive end -user
experience by enabling dynamic options such as drill-down and drill-through,
Dashboarding, Ajax controls and actions, and more.

About LogiXML
LogiXML, the leader in interactive, Web-based Business Intelligence, empowers
enterprises to turn data into business-critical information with pure Web-based reporting
and analysis products. The Company offers a comprehensive platform that addresses all
key areas of BI—managed reporting, ad hoc reporting, analysis and data services. Used
by thousands of organizations worldwide, LogiXML products are built on standards-based
technologies for easy integration, implementation and upgrade. LogiXML’s per server

 2009, LOGIXML, INC. PAGE 9


LOGI XML W HITE PAPER | A NEW MODEL FOR CORP ORATE SUCCESS

pricing model makes its powerful technology the most affordable BI solution on the
market. Founded in 2000, LogiXML is privately held and based in McLean, Virginia.

C o n t a c t U s T o d a y!

To set up a free live demo via the Web or to learn how the Web-based Logi 8 BI platform
can meet the BI needs of a broad spectrum of end-users across your company, please
contact us at: :

LogiXML, Inc.
7900 Westpark Dr., Suite T107
McLean VA 22102

TEL 1.888.LOGIXML | 703.752.9700


FAX 703.995.4811
WEB http://www.logixml.com
EMAIL sales@logixml.com

LogiXML and Logi Intelligence Server are trademarks of LogiXML, Inc.

 2009, LOGIXML, INC. PAGE 10

Potrebbero piacerti anche