QlikView Architecture Overview

QlikView Architecture Overview
QlikView Component Architecture
In-Memory Technology
1. Easily consolidates multiple data sources
2. Loads all data into memory, not just
selected cubes or static data sets
3. User selections (queries in traditional BI)
only initiate traffic between RAM and CPU
The Associative Experience
Traditional
Associative
Patient
Patient
Location
Episode
Location
Episode
Visits
IT-Driven
• Linear, pre-defined thinking
• Insights missed in hidden data
• Months to change
• Data-centric
Visits
User-Driven
• Follows the user
• All data is always visible
• Minutes to change
• Insight-driven
QlikView Skills Are a Great Fit
.QVW Files
A .QVW file is the application file for QlikView. It contains the scripting, data
model, data and user interface for a dashboard (or application).
ODBC
Script
OLEDB
XML
Data
Custom
Data
QlikView Desktop
File
DATA
GUI
.QVW
Security
SAP
SFDC
QlikView
Developer
QlikView
Application
File
QlikView is a Platform That Scales
10,000X
IE plug-in
AJAX
Enterprise:
NUMBER OF USERS
25X
QlikView Local Client
Clustered QlikView Servers
Clustered QlikView Publishers
iPad
iPhone
10X
Departmental:
QlikView Local Client
QlikView Server
QlikView Publisher
1X
Small Workgroup:
QlikView Local Client
Single User:
QlikView Local Client
(Personal Use License)
DEPLOYMENT TYPE
Blackberry
Android
.QVD Files
A QVD (QlikView Data) file is a file containing a table of data that QlikView has
extracted from one or more data sources. QVD is a native QlikView format and
can only be written to and read by QlikView. They are created with the scripting
that is included in the QVW files.
Benefits:
• Single Source of Truth
• Resource Flexibility
• Development Flexibility
• Development Speed
• Delivery Flexibility
• Incremental Loads
• Very Fast Data Loads
QlikView is an Architecture That Scales
QVD
Generator
App
QVD
Generator
App
QVD Xform
App
QVD
Generator
App
Co-Development - Utilizing Existing Skills
I.T. or Central
QlikView Team
Interface Designers
(business unit or I.T.)
Traditional BI Skills
• Querying
• Data Modeling
• Data Analysis
Design Skills
• Drag-drop
• Menu driven
BI Delivery Quadrants
enable these
while you
work on these
• Data Discovery/Scoping – QlikView developers mash-up data sources to determine certified
sources and scope requirements
• QlikView Projects – Develop from sources to QVDs to QVWs
• Business Discovery – allow enabled designers to “discover” in a sandbox or safe environment
• Self Service – allow collaboration and designer users to design from existing QlikMarts or templates
Enterprise Case Study – Small Deployment
• SAP Data
• Oracle Data
• 16 QV Data Files
• 1.5 GB Data
• ~100 M rows
• 14 loaded in full – daily
• 2 loaded incrementally
• 8 Dashboards
• Reloaded Daily
• Avg. 2 minutes load
Publisher
Just 4 moving parts
QVD File
QVW File
12 CPU cores
64 GB RAM
• 240 users
• 150+ daily
QV Production Server
Server
Publisher
Enterprise Case Study – Mid-Sized Deployment
• AS/400 Orders
• Oracle Data Mart
• SQL Server DBs
• Excel Files
• 84 QV Data Files
• 12 GB Data
• ~800 M rows
• 58 loaded in full – daily
• 26 loaded incrementally
• 23 Dashboards
• 21 Reloaded Daily
• 2 Reloaded Weekly
Publisher
8 CPU cores
32 GB RAM
Just 4 moving parts
QV Test/Dev Server
16 CPU cores
64 GB RAM
QV Production Server(s)
QVD File
• 600+ users
• Laptops
• iPads
• iPhones
QVW File
Server
Publisher
Enterprise Case Study – Large Deployment
• SAP Data
• Oracle DW
• SQL Server DBs
• Flat Files
• 148 QV Data Files
• 62 GB Data
• ~2.8B rows
• Full and incremental loads
• ~2 hours load time per night
Publisher 1
(QVDs)
• 52 Dashboards
• Daily & Hourly Loads
• ~1 hour load time per night
• ~ 5 minute load time per hour
Publisher 2
(QVWs)
16 CPU cores
48 GB RAM
QV Dev Server
Just 4 moving parts
QV Test Server
QVD File
QVW File
• ~4,600 users
• Laptops
• iPads
16 CPU cores
128 GB RAM
QV Production Server(s)
Server
Publisher
Scalability Example
One QlikView
dashboard can do the
work of dozens of
traditional BI
deliverables…at a
fraction of the cost.
Dimension Tables
· Customer
· Sales Rep
· Region
· Sales Manager
· Stocks
· USD Rates
· Item Master
Total Rows: 126,000+
Fact Tables
· Sales
· Budgets
Total Rows: 52,000,000+
User Interface
· 9 metrics sheets
· 47 charts
· 16 list boxes
· millions of filter combinations
· ALL in memory
Development
· 9 SQL queries pulled from
existing PL/SQL reports
· 2 weeks development total
· minimal IT help needed
Performance
· App open: 3.5 secs
· Avg filter response: < 2 secs
· # of database calls: 0 (zero)
· Backend network traffic: 0 (zero)
· 310 concurrent user capacity*
* based on single 64 GB RAM
quad-core server
QlikView Metadata
3 Types
• Descriptive – provides rich context about the makeup of a document
• Administrative - Provides centralized or application-specific views of
application reloads, user access, usage, performance, and scheduling
• Structural - Describes elements of an application such as its data sources
and repositories, tables, columns, expressions, charts, and graphs
Collection
• Can be collected from QlikView objects at any time. Does NOT
need to be forced on developers and designers as they construct.
• QlikView provides a “MetaScanner” (QlikView) application to
accomplish this.
Storage
• Stored in a QVD structure – simple, efficient and fast
• Easily readable or exportable to other formats – embed in dashboards
or use monitoring tools to explore the metadata in QlikView
• Uses QlikView to Manage QlikView – no additional software/hardware
Speed
Sandbox
Collaboration
Co-Development
Self-Service BI
Scalability
Capacity Plans
Monitoring Tools
Meta Model
Single Source of Truth
QlikView Enterprise Framework
• best practices
• white papers
• demo apps
• code samples
• tips and techniques
…..for enterprise scaling and deployment of QlikView.
High Level Security Overview
QlikView Security – Dynamic Reduction
QlikView Security – Loop and Reduce
Server Components and Scaling
QlikView Server - Components
QlikView
Developer
QlikView Server - Three-tier Architecture
• To create a tree-tier Architecture put
the QlikView Web Server or IIS on a
separate front server (Virtual machine)
in front of the QlikView Server.
• This design works better with AJAX
only. The QVP protocol will
communicate directly to QlikView
Server when using the Plug-in Client.
Hardware Scaling – Scaling Up QlikView Server
• Add more memory
• Add more CPUs or cores
Hardware Scaling – Scaling Out QlikView Server
• Add additional servers to create a
QlikView Server Cluster
• Automatically loading applications
on each node for maximum memory
balance
• A hardware load balancer or DNS
round robin could be added for http
failover
• QVWS/IIS is on the same server as
the QVS
• No Windows cluster functionality is
needed
Hardware Scaling - Scale up QlikView Publisher
• Add more memory
• Add more CPUs or cores
Hardware Scaling - Scale Out QlikView Publisher
• Add additional servers to create a
QlikView Publisher Cluster
• Publisher tasks will automatically
distributed between the nodes
according to a configurable memory,
load and CPU formula.
• No Windows cluster functionality is
needed
Scaling example in a Three-tier Architecture
• Adding extra web servers in the
presentation tier creating a QlikView
Web Server cluster
• Use a load balancer on top to balance
HTTP/HTTPS traffic to the nodes
• Adding extra QlikView Servers in the
Application tier
• Creating QlikView Server cluster
• Or two separate QlikView Servers
User License CALs
Thank You!