Data Warehouse Like a Tech Startup with Oracle Autonomous Data Warehouse

© MJR Analytics 2018, T: +44 01273 041134 (UK) 415-218-2161 (US) W: https;//mjr-analytics.com E: info@mjr-analytics.com
Data Warehouse Like a Tech Startup with
Oracle Autonomous Data Warehouse Cloud
Mark Rittman, CEO and Founder, MJR Analytics
Oracle Open World 2018, San Francisco

Introducing MJR Analytics
● Specialists in Oracle Cloud Analytics
● Founded by Mark Rittman in 2018
● 100% Cloud focus + project delivery
○ Oracle Analytics Cloud
○ Oracle Autonomous DW Cloud
○ Oracle Data Integration Cloud
○ Oracle Big Data Cloud
● Speak to us now during OOW 2018
info@mjr-analytics.com
+44 7866 568246
https://www.mjr-analytics.com

It’s Good to be back at Oracle Open World
● Former co-founder and CTO of an Oracle BI+DW consultancy
● Oracle ACE Director, Author of two Oracle Press books
● Consultant specialising in architecting, building and tuning Oracle BI+DW
● Regular speaker at every Oracle Openworld since 2006
● … but then in 2016 I took two years out

I joined a tech startup

Agile self-organizing teams
Thinking product not consulting
Real-world big data vs. PoCs and BigDataLite
Monthly product releases

And everything I thought I knew, was wrong

Product Manager for Platform Analytics Features
Raw, Event-Level Tables optimized for fast
ingest and processing
Industry-Vertical Business Views
Customer-Specific Semantic Model
Materialized Historic
View Rows
Real-Time Data View
(Today’s Data)
Base Views Industry Extensions
Base + Industry Explore + Extensions

Startups need to move fast.
“The battle between every startup and incumbent comes down to whether the
startup gets distribution before the incumbent gets innovation.”
- Alex Rampell, A16z, “Distribution vs. Innovation”

This startup was built on data
Lots of data

150,000+events/second
>2PBdata in 70,000+ tables
1Bn+interactions/day

zero DBAs

They didn’t have time to take backups

They’d never heard of an index

They didn’t do server patchsets.
Or servers.

Or own any hardware

So what did they care about?

Agility and time to provision
Scaling their capacity up (and down)
Costs that start small, and align with growth

Scale-out solves one problem, but then you need to
manage that cluster of servers
Data Warehouse RBDMS
> 3 nodes, >1 TB
Hadoop / HBase cluster
> 50 nodes
1000+
HBase regions per server

Elastic Compute + Storage
Elastic Pricing

MVP - “Adoption or fail fast”
GA Release
Initial Early Adopters
Product / Market Fit
Minimal cost
Pay via CC and expense
Costs linear with product
adoption
Can get expensive

No-Ops Autoscaling
Black Friday
New Year Sales
De-provisioning
of excess nodes
-$$$
Provisioning
of additional nodes
+$$$

Elastic Costs based on Planned Workload
Black Friday
New Year Sales
Automatic provisioning
of additional nodes
+$$$
De-provisioning
of excess nodes
-$$$

No-Ops Autoscaling
Black Friday
New Year Sales
Automatic provisioning
of additional nodes
+$$$
Automatic
de-provisioning
-$$$
So you don’t waste $$$ on capacity you don’t need

Startups need to focus.

Serving Customers
Building Products
Domain Knowledge
Patching servers
Doing backups
Scaling capacity
Things you are good at Things you suck at
Do These Automate this

Data Analysis within a Tech Startup

Data Warehouse Analysis within Startups
Provision a Cloud
DW Instance
Land Data in
Object Store
Analyze via BI
Connector
Deprovision
Cloud DW
Instance
Create simple
column-store
tables

How do these database services work?

However...

If you need to know, they’ve failed

But they do have limitations

Limited or non-standard SQL
Primarily INSERT-only
Distributed joins + resource limits

Coming to Disrupt Your IndustryNow you can do the disruptingOracle Autonomous Data Warehouse Cloud

Oracle Autonomous Data Warehouse Cloud
● Fully-managed DW Platform-as-a-Service
● Based on Oracle 18c Exadata Database technology
● Near-instant provisioning
● Elastic scaling and pricing
● Simplified column-store table creation
● Automated provisioning, patching and upgrades
● Automated backups
● Includes data visualization + notebook apps

Data Warehouse Analysis using ADWC

Provisioning an Autonomous DW Cloud Instance
● Provision from https://cloud.oracle.com
● Choose DB name, display name
● Select CPU count (1 min) and storage (1TB min, 128TB max)
○ Creates tenant within multi-tenant Exadata instance
○ Scale-up and down within 128 CPU/TB instance limit
● Provisioning typically takes ~ 1 min
● Download client credentials as secure wallet ZIP file

Loading Data into Autonomous Data Warehouse
● Data for analysis can uploaded as part of table creation
● More typically, data is located in OCS/S3 object storage
○ Allows sharing of data with other cloud applications
○ Real-time analysis of data streamed via Event Hub

Data Lake on Oracle Big Data Cloud Topology
● New DBMS_CLOUD package
for object store operation
○ Analyze OCS/S3 data
in-place as external tables
○ Or load into Autonomous
Data Warehouse Cloud
● Extends scale of ADWC beyond
limits of Exadata storage

Simplified Table Creation & Accessing Data
● “Load and go”
○ Define tables, load data, run queries
■ Typically using free SQL*Developer GUI
○ Because it’s based on Exadata technology
■ No indexes
■ No need for partitioning
■ No need for materialized views
○ Because it’s Autonomous
■ Self-tuning and self-optimzing

Simplified Table Creation & Accessing Data
● Comes with entitlement to DV Desktop license
● Includes web notebook based on Apache Zeppelin
● Connect most popular BI + data integration tools, e.g.
○ Looker

Scale-Up and Scale-Down of Compute + Storage
● Ability to scale-up / scale-down with steps then automated
○ Pricing based upon CPU ($/CPU/hour) and
Storage ($/TB/Month)
● Close your ADWC instance when not needed, restart instantly
○ Leave data in ADWC storage and set CPU to 0
○ CPU and storage to 0 and data left in object store

Scale-Up and Scale-Down of Compute + Storage
● Independently scale compute or storage
○ Resizing occurs instantly, fully online
○ Memory, IO bandwidth, concurrency scales linearly with CPU
● Automates resizing tasks - not autoscaling (yet)

So why Autonomous Data Warehouse?

Move like a startup without
introducing new technology
to your mix.

Elastic scaling and elastic pricing
… but now using a real database

https://www.forbes.com/sites/oracle/2018/10/04/how-a-big
-business-can-use-autonomous-it-to-move-like-a-startup/#15b
9df6e1752

MJR Analytics for ADWC
● Specialists in Oracle Cloud Analytics
● Founded by Mark Rittman in 2018
● 100% Cloud focus + project delivery
○ Oracle Autonomous DW Cloud
○ Oracle Data Integration Cloud
○ Oracle Big Data Cloud
● Speak to us now during OOW 2018
info@mjr-analytics.com
+44 7866 568246
https://www.mjr-analytics.com

Data Warehouse Like a Tech Startup with Oracle Autonomous Data Warehouse

More Related Content

What's hot (20)

Similar to Data Warehouse Like a Tech Startup with Oracle Autonomous Data Warehouse (20)

More from Rittman Analytics (13)

Recently uploaded (20)

Data Warehouse Like a Tech Startup with Oracle Autonomous Data Warehouse