1

Mandate or Option?Analytics in the Cloud

2

Buy Any WayEmpowers companies to purchase software

in more accommodating ways based on specific use cases through simplified pricing bundles, subscription-based licenses, and

as-a-service options

Analyze AnythingEnables analytic users throughout the

organization to use their preferred analytic tools and engines across data sources, at scale

Move AnytimeFuture-proofs buying decisions by taking

advantage of our software license portability that provides flexibility to run analytics across

deployment options

Deploy AnywhereProvides analytic processing across flexible

deployment options, including the Teradata Cloud and public clouds, as well as on-premises on Teradata hardware or commodity hardware

Teradata Everywhere™

3

What is the Cloud?The cloud is a model, not a destinationCloud computing is a model for enabling convenient, on-demand network access to a shared pool of computing resources that can be rapidly provisioned and released with minimal management effort or service provider interaction.

According to the National Institute of Standards and Technology (NIST)

Essential5Characteristicsof CloudComputing

1. On-demand self-service

2. Broad network access

3. Resource pooling

4. Rapid elasticity

5. Measured service

Provisioning of computing capabilities (compute, network, storage) without human interaction.

Services are available over the network and accessed through standard mechanisms.

Computing resources are pooled to serve multiple consumers using a multi-tenant model.

Capabilities can be elastically provisioned and released to scale rapidly.

Resource usage can be monitored, controlled, and reported.

Why are you thinking about moving to the

Cloud?

If you don’t have a good answer to this question, you won’t be able to

measure success.

6

Why Companies Choose the Cloud?

Agility Usability Scalability Cost Modernization

7

Common Misconceptions About the Cloud

It’s ComplicatedMigrating to the cloud needn’t be complicated if you break down the process. A good adoption plan uses cloud services that easily integrate with existing on-premises systems and security models. Choosing the right service model can further simplify the transition.

Its’s Not SecureCloud providers focus on security and can leverage economies of scale not available to individual organizations. In fact, Gartner predicts that by 2020, public cloud infrastructure workloads will experience 60% fewer security incidents than those in traditional data centers.

It changes workflows and processesDetermining which applications make sense to move and which don’t on a case-by-case basis can minimize disruption and potentially eliminate noticeable differences in how employees access services after they move to the cloud.

8

Barriers to Cloud Adoption

4CommonAnalyticalUse Cases for the

Cloud

Analytic SandboxesUser provisioned workspaces that allow business people to conduct discovery and situational analytics. Ideal for power users who need more flexibility

Seasonal CapacityExtra capacity when you need it to handle peaks in data requirements

Inexpensive Data StorageStorage for large quantities of IoT data, web logs and other data that has yet to be analyzed to determine its value to the organization

Disaster RecoveryBackup systems that need to be physically located separated from primary systems for use in emergencies

10

4 Service Models for Cloud Computing

Infrastructureas a Service

Platform as a Service

Softwareas a Service

Refers to vendor-provided hardware and software.

Example: Teradata Cloud, OpenStack, AWS, Microsoft Azure

Set of tools and services designed to make coding and deploying those applications quick and efficient.

Example:Teradata Managed Cloud, AWS, Microsoft Azure

Includes standalone Web applications or thin client access that support various computing functions.

Example:Gmail, Salesforce, SAS

Resultsas a Service

No software license or infrastructure to buy. You provide the data and the business problem and a third-party delivers the results.

Example:Uber, SAS

11

Illustrating the Cloud Computing Models

Infrastructureas a Service

Platform as a Service

Softwareas a Service

Resultsas a Service

Private CloudCloud services are only available to members of a single organization. The infrastructure can be hosted by the same organization or a third-party

Public CloudCloud services are offered to the general public. The infrastructure is hosted by anyone.

Hybrid CloudComposition of two distinct infrastructures with one being a cloud, such as a Private + Public cloud or a Cloud + On-Premise combination.

3Types of

DeploymentsCloud

Criteria for Evaluating

CloudServices and

Platforms

• Data Gravity

• Data Migration & Movement

• Scalability

• Backup & Restore

• Availability (Service Level Agreements)

• Network Connectivity

• Latency

14

The Hidden Costs of the Cloud to Avoid• Application Charges• Data Transfer Charges• Extra or Unused Resources

• Data Storage• Applications• Network Resources

• Underutilized Discount Capacity• Poorly Selected Storage Types

15

Transitioning to the CloudThe cloud is not an all or nothing proposition!Companies need to carefully consider how, when and what they will move to cloud. Not all applications make sense and some make more sense than others. Choose wisely and don’t fall victim to corporate mandates or directives.

16

Analytic Cloud Optimization in the Real World

AnalyticSandboxes

AnalyticData

Streams

On-PremiseData Sources

DisasterRecovery

IT Admin

Data Scientists Analytic Consumers Executives

Cloud Bursting forSeasonal Capacity

Data Transfer

Cloud SystemsOn-Premise Systems

InexpensiveData Storage

Focus on Minimizing Data MovementData movement is the largest bottleneck in the analytic lifecycle… on-Premise or in the cloud. Designing a system that minimizes or accelerating this movement is essential to improving analytic performance regardless of location

3 ways to minimize or accelerate data movement:

• Co-location of the data and analytics

• In-Database technologies

• High-speed data connectivity

17

Co-Location of Your Data and AnalyticsIt may seem obvious but minor oversights can cause big delays in analytic performance

Optimal Suboptimal

Single Cloud On-Premise to Cloud Cloud-to-Cloud

18

In-Database Analytics

• Data movement is minimized by pushing the analytics to the data

• Ideal for cloud-to-cloud or on-premise to cloud deployments

• Data can be lifted into memory for Discovery system done over high-speed BY/NET

• No need to sample data for discovery, model development or scoring

• Data duplication is minimized or even eliminated

Minimizes the need to move data by leveraging the processing power of the database

Data Warehouse

Model Development

Model Scoring

Analytics Movedto the Data

19

In-Database Analytics Supported for the use with Teradata in the Cloud

• SAS 9.4M4 extends support to Teradata in-Database Analytics running as a service in the following cloud platforms:• Teradata Database on Amazon Web Services – Public cloud• Teradata Database on Microsoft Azure – Public cloud• Teradata IntelliCloud – Teradata managed cloud• Teradata Database on VMware – Private cloud

• Based on the work performed by SAS R&D and Product Management, these system will function the same as an on-premises Teradata databases

SAS Expands Support to Teradata Database as a Service on Cloud Platforms

20

Not All Clouds Perform the Same

Teradata’s database software is better for advanced analytics

• An 8 node Teradata system outperformed Hadoop and a leading cloud database with 32 nodes each*

• Teradata’s software running on AWS was able to perform 17,000 more queries per hour. The on-premise solution does even better.

• More queries per hour equates to:

• Lower overall system cost

• Improved analytic performance (faster analytics)

• The ability to more complex analytics

17,000X

* All 5 systems used the same 24 Terabytes of data and all ran the same mixed workload queries – 50 streams of analytic and 300 streams of tactical

21

For more information, contact us at sas@Teradata.com

Thank You!