Aug 2012 HUG: Hug BigTop

37
What's stacking? Or: How I Learned to Stop Worrying and Love the BigTop Dr. Konstantin Boudnik Apache Hadoop committer since 2009 Co-author of iTest (BigTop's kernel) Senior architect @Karmasphere [email protected]

Transcript of Aug 2012 HUG: Hug BigTop

What's stacking?

Or: How I Learned to Stop Worrying

and Love the BigTop

Dr. Konstantin BoudnikApache Hadoop committer since 2009

Co-author of iTest (BigTop's kernel)Senior architect @Karmasphere

[email protected]

Distributed under CC3.0https://creativecommons.org/licenses/by/3.0/

Disclaimer!

All characters appearing in this work are fictitious. Any resemblance to real creatures, living or dead, is purely

coincidental

Can you see the forrest behind the trees?

Not a technical talk...

What's the real intrinsic value of BigTopWhat is our philosophical proposition

Why Hadoop data stack is important?

The elephant is yesterday's news...

An article over at Forbes:

“...it is not a database, it is not an analytics environment, and it is not a visualization tool. By itself, it is not a solution for helping businesses make better decisions.”

Some self-proclaimed hot-shot dude

Sorry, I am allergic to bullshit...

There are limits, all right...

● Maximum size of the file system● High availability● Recoverability

But there is a lot to pick from...

● Do you want to experiment with– CEPH

– Cassandra

– Ganglia

– Puppet

Why I am wasting your time?

● You are an ASP and need to ship a stack

Why I am wasting your time?

● You are an ASP and need to ship a stack● You need a higher level abstraction for MR

Why I am wasting your time?

● You are an ASP and need to ship a stack● You need a higher level abstraction for MR● You are in need of virtual deployment

Why I am wasting your time?

● You are an ASP and need to ship a stack● You need a higher level abstraction for MR● You are in need of virtual deployment

Hey!

“I am gonna just pick up a distro from B or S”1)

1) “B” - Big Ego Distributor “S” - Small Ego Distributor

Yeah, right! (case study)

Commercial vendor “BS” packs Hive 0.7.1 into a release

We know how it is full of it

Good news!

“Our next release comes with Hive 0.8.1”BS' VP of Product

But may be they just forgot this?..

public int executeUpdate(String sql, String[] columnNames) throws SQLException {

throw new SQLException("Method not supported");

}

So, no worries! We can deliver!

But can they?

Do you REALLY have to be at mercy of a distribution vendors?

“I have no resource to do it myself”

DIY: and there's always “but”...

Maybe ASP or ISV business isn't for you?

Or maybeBigTop stack: take it and go...

Modify a stack BOM– Build

– Deploy

– Configure with Puppet (included)

– Test (scenarios are provided or easy to add)

– Grab a virtual appliance if short of hardware

Hit a problem?Just rinse and repeat (C)

Many thanks to the good folks from Hadoop community!

● Many components● Gazillions of versions● Lot of patches if you like it hot and dirty

Where is the philosophy?

WAIT!

Competition is great!● Competition enabling open platform● Diversity of stacks● No more vendors lock-ups● Don't be afraid to try● Be your own Sprechstallmeister

May be something wrong with this?

News And Updates

5 releases of BigTop!

Three are out Two more in the work

Strong community and growing!

20 committers and many contributorsWe are looking for more!

And sorry: no more binaries for you

BigTop won't be publishing convenience binary artifacts anymore

Might as well use this expertise:http://www.magnatempusgroup.net

Come and join us!

Powered By Bigtop:ClouderaEMCHortonWorksMagnaTempusGroupTrend Micro

https://incubator.apache.org/bigtop/

https://blogs.apache.org/bigtop/

May be one day...

Q & A