Architecture Modernization with Cloudera

[pic 1]

[pic 2]

[pic 3][pic 4][pic 5][pic 6]

Architecture modernization

Dear Attendees

Architecture modernization

Who doesn’t want want to work with the latest architecture ?

But who knows what this is ?

After our presentation you will know better why you should modernize your architecture, are you more aware of differetent tools available and .. and will you get a better insight in things to do

http://jalopnik.com/the-tesla-model-s-raises-the-suspension-based-on-locati-1636877061

Bob, Jim Here . we got a lot of complaints about over damaged Battery covers and scratches at the bottom of the car

Could your team have a look at it  ? I Don’t wont to a recall

Next day

Jim, we found some strange stuff. Every car that has a damaged cover has unusual sensor activity on certain days and locations. Especially the air suspensions and shock absorbers reacted heavily

We looked deeper into the location data and found that a lot times it happened on off road tracks.

The idea of my team is to lift the car automically 1. 2 inch  when the  sensors react heavily.

We can build new air suspensions settings  into the next software release

My Name is Frank Vullers , I am a Business Value Consultant for Cloudera. I helps customers with finding use cases (discovery workshops) and calculating the benefits (value assessments).

Our relationship with data is changing.  In the old days was data the result of an action. Now data it self can be an action

TRANSITION: Lets have have a look at how data is used

Evolution in use of data

We use a lot of data in traditional Business Intelligence. Daily reports etc

In todays world more and different data is now available. With Big Data Analytics we have new possibilities

The world goes faster and faster and also BI needs to react. Fast Data Analytics is the answer to this

In the next slides I would like to go deeper on these 3 areas

Traditional BI

In the classic bI world , the business starts with asking questions , IT structures the data to answer the questions.

Data is captured only what is needed

The traditional Analytics like statistical analysis of  or segmentation is used to come with answers

Who are our customers?  What do my customers segments spend etc

Big Data Analytics

Big data analytics works in a different way.  All data is captured in case it’s needed, multistructured. Business explores the data to find questions worth answering

Who will be our customers in 6 months and where will they come from?

In this world new analytics is used like

• Path analysis (eg to see how customers move from gold to silver to bronze customer card over time
• Text analytics; finding out how people react on your commercials via reading tweets
• Graph Analytics eg find which customer is most influential and should be rewarded
• Map reduce

Fast Data Analytics

More and more we want to act on real time/ near real time events. First we need to do the analysis in the given time frame, next we have to react.

Examples of this are

Event Detection

• Fraud/Risk Detection
• Spam Filter
• Marketing Alerts

Recommendation Engine

• Next Best Offer
• Content and/or Services Recommendation

Model scoring

• Embedded Analytics
• Analytic Aggregates
• Reports

TRANSITION: Now we have seen the changes in Data usage it is time to switch to architectural views

Architecture View

Lets go to the architecture view

Schema on read is the change agent

An important aspect of the enterprise data hub is support for both 'schema on write' and 'schema on read' in order to handle routine and exploratory workloads.

• Schema on write (as with traditional databases) provides good performance as it is possible to lay out the data efficiently, as well as good governance.
• Schema on read allows users to store any data as the system looks more like a file system than a database. It effectively performs ETL (extract, transform, load) on the fly at read time, generating the appropriate schema as part of the process. This means an additional column of data can be provided for analysis very quickly.

The logical architecture hasn’t changed

We see here the logical architecture of Datawarehousing . Ralph Kimball talked about it in the seminar The Future of data ware housing. ETL will never be the same.

I don’t want to go into the details of this architecture but highlight the fundamental differences.

Ralph talks about the EDW backroom and the EDW frontroom

We see data coming in from the orginal source systems, being processed  in the ETL step  

, Data exposed into the the presentation layer and  the BI applications

BUT, the physical architecture of the back room now looks very different

When we bring our Enterprise Data Hub into the game we bring in HDFS files and most imprortant Schema on read

Old Backroom