Pentaho Business Analytics Evolves Pedro Alves Pentaho SVP Community & Product Designer, Hitachi Vantara
Agenda • Pentaho Vision • High Level Roadmap • Pentaho Analytics – Yesterday • Today: The Beginning of a Journey
Pentaho Vision
A product vision must be a combination of 3 elements
What we want What What the market we have wants
Top down approach: Strive for something the users desire
Usability first
We won’t improve the product usability if we think about technology
We have to think differently
So instead of thinking about Datasources, APIs, ETL, Models, Dashboards, Architecture…
Lets think about what the users want
They want information
Pentaho Vision – Making it simple Pentaho Proprietary and Confidential
Pentaho Vision: The story of an insight Pentaho Proprietary and Confidential
No more separation between DI and Analytics Data Preparation Pentaho Proprietary and Confidential
Analytics must be available anywhere Data Preparation Pentaho Proprietary and Confidential
And collaboration is fundamental Data Preparation Pentaho Proprietary and Confidential
This insight then comes to life Collaboration Pentaho Proprietary and Confidential
A library of insights, In a governed, scalable and collaborative universe Sharing and consumption Pentaho Proprietary and Confidential
And consumed any way needed Different outputs Self Service Dashboard Visualizations DET Self Explore CTools Dashboard Report Pentaho Proprietary and Confidential
How are we executing on this vision?
High-level Roadmap
Message: A Single Flow with Analytics Anywhere Pentaho Proprietary and Confidential
From Vision to Roadmap (Big) Data Visual Data Enterprise Big Data Processing Platform Experience Processing Data Explorer Spark Engine Pentaho Scale-Out Visual Data Prep Adaptive Execution Cloud Deployment Streamlined Modeling Streaming Lumada Integrated Data Science Extensions Machine Learning
Visual Data Experience Visual Data Big Data Experience Processing Data Explorer Visual Data Prep Streamlined Modeling Data Science Extensions Pentaho Proprietary and Confidential
(Big) Data Processing (Big) Data Processing Spark Engine Adaptive Execution Streaming Machine Learning Pentaho Proprietary and Confidential
Enterprise Platform Enterprise Platform Pentaho Scale-Out Cloud Deployment Lumada Integrated Pentaho Proprietary and Confidential
Pentaho Analytics: Yesterday
Analytics have always been relegated to the Pentaho Server
On a tool based approach
That we chose in the first place
Pentaho Server Analytics Tools Interactive CTools Pentaho Analyzer Big Data Reporting Dashboards Dashboards Processing Pentaho Proprietary and Confidential
With different and sometimes overlapping capabilities
Pentaho Server Analytics Tools Interactive CTools Pentaho Analyzer Big Data Reporting Dashboards Dashboards Processing Pre-canned Dashboard Self Service Ad-hoc Pivot View Embedded capabilities Template based Ad-hoc Row Level analysis Visualizations Allows the Art of the possible Limited set of components Metadata Based Inline Modelling Easy to use Not easy to use Report generation Several functions Pentaho Proprietary and Confidential
Today: The Beginning of a Journey
In order to fix usability, we had to take a step back
The first ask can’t be: What tool do you want to use?
Analyzer is based on the Mondrian engine
Visualizations were tied to it and not reusable
Interactive Reporting is based on the Pentaho Metadata engine
Both require separate modelling exercises
And don’t share common functionality
Don ’ t get me wrong…
They are both excellent tools for the use cases they were created for!
But we pay the price for maintaining two different applications
And users pay the price for the lack of coherency
We want to fix it
We want to fix it. And we will!
So we’re aiming for 2 objectives:
1. A unified and coherent way to explore data
2. That can be used anywhere in the cycle
Internal Project Codename: DET
We envisioned a tool that abstracted all those concepts
And could, in time, provide the features each existing tool provides
And be the place to add new capabilities
Such tool wouldn’t be built from scratch…
Instead, would leverage existing technology
But hiding it from the end user
They don’t care about models, engines, any of that crap…
Now…
Some of you may not know…
But a first version of this tool already exists since 7.0
If you haven’t seen it, you’re looking in the wrong place
It’s not here…
It’s here
We’re building something that allows users to explore any data
No matter where it comes from
And giving them the freedom to look at it any way they choose from
At a row level…
As a Pivot table
Or any other visualization
Since 7.0, we’ve been working on several threads:
Increased functionality Drilldown Filters Actions (Keep only / Exclude / etc)
Extensibility and coherency Viz Api 3.0 Support for Analyzer and CTools
Performance improvements Reducing start-up time Optimizing resources Handling more data
And going forward we’ll keep working on many more
More functionality
Inline modeling capabilities
Profiling capabilities
Support for bigger and more complex datasets But thinking about this differently….
Write-back capabilities (Wrangling) The real PDI thin client strategy, a wysiwyg ETL layer
Ability to plug in your own data Always in a governed way
Available everywhere: PDI, Pentaho Server…
Or embedded in your own application
Pentaho Analytics: Tomorrow
Going back to the original vision
We want to remove the focus from the tool
And put the focus on the information
The insights we get from them
And let everything else follow… Visualizations Self service ”dashboards” Printed reports Custom Dashboards Blending on the glass
Allowing us to implement the original vision
Pentaho Analytics Self Service Dashboard Visualizations DET Self Explore CTools Dashboard Report Pentaho Proprietary and Confidential
Recommend
More recommend
