Big Data Shouldn't Be Big

22
Big Data Shouldn’t be Big

Transcript of Big Data Shouldn't Be Big

Page 1: Big Data Shouldn't Be Big

Big Data Shouldn’t be Big

Page 2: Big Data Shouldn't Be Big

ZoneFox – Who Are We ?

• Spin-out from Edinburgh Napier University

• Endpoint Detection and Response

Page 3: Big Data Shouldn't Be Big

ZoneFox as Big Data

• ZoneFox streams and collects endpoint data

• On average we write more than 300 million events per

day, per client

• 1 Trillion events on average for a 500-endpoint

customer

Page 4: Big Data Shouldn't Be Big

How do our Customers use ZoneFox?

Page 5: Big Data Shouldn't Be Big

What we learned

• Theoretical

• Practical

Page 6: Big Data Shouldn't Be Big

In Theory…

Page 7: Big Data Shouldn't Be Big

“Big data is a term for data sets that are so large or

complex that traditional data processing

applications are inadequate.”

Page 8: Big Data Shouldn't Be Big
Page 9: Big Data Shouldn't Be Big

Vs.

Page 10: Big Data Shouldn't Be Big

• 332 million active users

• 300-500 million tweets per-day

• 1.65 billion active users

• Every 60 Seconds:

– 510 Comments

– 293,000 statuses

– 136,000 photos

Page 11: Big Data Shouldn't Be Big

Source: Pew Research Center Source: Cowen & Company

Page 12: Big Data Shouldn't Be Big

$250 Billion$10 Billion

Page 13: Big Data Shouldn't Be Big

What does Facebook get right?

• Meaning

• Focus

Page 14: Big Data Shouldn't Be Big

Meaning

• Data is not Information

– Relevant vs Irrelevant

– Signal to Noise

Page 15: Big Data Shouldn't Be Big

Focus

• Volume

• Velocity

• Variety

= Value

Page 16: Big Data Shouldn't Be Big

In Practice…

Page 17: Big Data Shouldn't Be Big

ZoneFox – Architecture

Endpoint Agents [Windows, Linux &

SQL Server]

Collector & Presentation

Server(s)[Microsoft Windows]

Database, Reporting & Search Server(s)

[Ubuntu]

Horizontally Scalable. Proven to handle 10,000+ endpoint agents

Secure, Highly Compressed

Messages

Lightweight, Zero-Configuration

Agent

Page 18: Big Data Shouldn't Be Big

ZoneFox Data

• Simple Event Model

– Machine

– User

– Process

– Object

– Behaviour

– Time

Page 19: Big Data Shouldn't Be Big

What Does this Mean?

• We’re able to tell when

– Bob in HR starts to act like Alice in Engineering

– Alice is thinking of leaving the organisation

– Eve uses shadow IT out of frustration

Page 20: Big Data Shouldn't Be Big

Rolling your own

• Technology is often immature, resulting in frequent release

cycles with breaking changes.

• Usually, 1 big data solution solves 1 niche problem

• Expect to trade some speed for consistency: getting results

faster means you have to compromise on data "freshness"

Page 21: Big Data Shouldn't Be Big

Rolling your own

• Very small talent pool, as most devs won't have much

experience with the latest technology X

• Security in big data is an actively developing area

• Any additional features are usually not available out of the

box.

Page 22: Big Data Shouldn't Be Big

Thanks – Any Questions?

Jamie Graves, [email protected]