VMware vFabric Data Director
Powering Database-as-a-Service for Oracle, SQL Server, Hadoop and
vFabric Postgres
Confidential
© 2010 VMware Inc. All rights reserved
Traditional IT
Application
Team
DBA
System
Admin
Storage
Admin
Network
Admin
Days - weeks
to provision a development environment
2
Database Infrastructure Challenges
Development
Production
Postgre
Test
Postgre
Postgre
UAT
Shadow IT
Postgre
Enterprise
DBA
Security
3
Compliance
Routine Tasks
vFabric Data Director Powers Database-as-a-Service for your Cloud
Enables database-aware virtualization on VMware vSphere and provides
database-as-a-service for heterogeneous databases and Hadoop
4
Reduce capex through
database-aware
virtualization
Increase IT agility
Save cost through
database and Hadoop
virtualization
Automate database and
Hadoop lifecycle
management
Accelerate analytics
and application
development
Through self-service and
automation
vFabric Data Director Platform Architecture
Enterprise Services
vFabric Data Director
Provisioning
Backup/
Restore
Clone
Database
Database
Ingestion
Catalog
One click
HA
DBA App Dev
Infrastructure
Services
Resource
Mgmt
Security
Mgmt
Template
Mgmt
Patch
Mgmt
Monitor
DBA IT Admin
…
vPostgres
Cloud
Infrastructure
Platform
Cloud
Infrastructure
Platform
5
.
.
.
Integration Gateway (API)
Lifecycle
Services
vCloud
Director
Application
Director
Cloud
Foundry
EMC Chorus
EMC
Data Domain
…
Supported Data Engines
 Oracle 10gR2, 11gR2
 SQL Server 2008R2, SQL Server 2012
 Postgres 9.1.6
 Hadoop 1.0 based distributions:
• Apache Hadoop 1.0.x
• Cloudera CDH3
• Greenplum HD 1.1 and 1.2
• Hortonworks Data Platform 1.0
• MapR - Soon
6
SQL Server Support - New
Data Director now supports SQL Server database
• Supports provisioning of SQL Server 2008 R2 and SQL Server 2012 databases
 Key features include
• Create SQL Server templates easily with built-in configurations
• Size VMs and configure database settings through a single UI
• Join VMs to domain as part of provisioning
• Compatibility with existing tools
7
Confidential
Provision Any Database from Same Web Portal
8
Provision and Manage Any Hadoop Distribution with One Click
9
Hadoop Support Powered by Serengeti
 Open source project launched in June, 2012
 To learn more - projectserengeti.org
Serengeti
Deploy a Highly Available, Elastic Hadoop cluster in 10 Minutes
Mix Hadoop with other workloads – leverage existing vSphere capacity
Scale Hadoop Compute on demand
Use Your Favorite Hadoop Distribution
One stop command center – configure and run Hadoop
10
Why Virtualize Hadoop?
Operational Simplicity
High Availability
Elasticity & Multi-tenancy
 Rapid deployment
 High availability for
entire Hadoop stack
 Shrink and expand
cluster on demand
 One click to setup
 Independent scaling of
Compute and data
 One stop command
center
 Easy to
configure/reconfigure
11
 Battle-tested
 Strong multi-tenancy
Data Director Benefits
Stretch Your Servers
Liberate Your Licenses
Double Your DBAs
Secure Your Stuff
12
Reduce Costs by > 50%
•
Consolidate servers by 10–15X
•
Save on Hardware, Software, Power and Cooling
Use the Right Database for the Job
•
Use the most cost-effective licensing for your application
•
Leverage low-cost HA and clustering with vSphere
Don’t Let Your DBAs Hit the 40 DB Ceiling
•
Automate provisioning and management
•
Enforce policies and best practices automatically
Keep Your Data Secure With Automation
•
Make security a holistic part of the database lifecycle.
•
Help admins keep systems patched and up-to-date.
Confidential
vFabric Data Director
Data Director will help you:
1. Stretch Your Servers
2. Liberate Your Licenses
3. Double Your DBA Productivity
4. Secure Your Stuff
13
Confidential
vFabric Data Director
Data Director will help you:
1. Stretch Your Servers
2. Liberate Your Licenses
3. Double Your DBA Productivity
4. Secure Your Stuff
14
Confidential
We Looked at 700,000 Servers
The average PRODUCTION Oracle Database
Instance uses:
• 4% of CPU
• 2MB/second of Network
• 1200 IOPS
The average SQL Server database only does
slightly better (about 5% CPU).
This does not strain a
modern server in any way.
15
Confidential
How Can We Do Better?
Virtual
Server
Physical
Server
75%
5%
0%
95%
20%
You can consolidate 10-15 databases
on a single server without affecting
overall
60%performance.
80%
100%
40%
Used
16
25%
Unused
Confidential
Overhead
vFabric Data Director
Data Director will help you:
1. Stretch Your Servers
2. Liberate Your Licenses
3. Double Your DBA Productivity
4. Secure Your Stuff
17
Confidential
How Can We Use fewer Licenses?
Virtual
Server
Physical
Server
75%
5%
0%
18
25%
95%
20%
Oracle and SQL Server are typically licensed
40%If you consolidate
60%
100%
by core.
by 80%
a factor of 10-15x
Used
Unused
Overhead
you
will need
far fewer
licenses.
Confidential
How Can I Virtualize a Physical Database
Data Director Application
DB Management Server
Shell Virtual Machines
Oracle 11.1 Template
Oracle 11.1
TEMP
Monitoring Agents
Linux
SWAP
T
Linux
Physical DB Server
Empty
DBVM DB
Clone
of Physical
Oracle 11.1
Monitoring Agents
Linux
19
Confidential
LOG
DATA
A Single License Management Pane
Don’t get a million dollar
surprise because you couldn’t
track your database licenses.
20
Confidential
A Real World Example
EMC is one of the largest Oracle shops in the
world and it runs on VMware.
EMC was able to:
• Consolidate their servers by a factor of 13:1
• Reduce RAC to Single Instance databases
• Convert Enterprise to Standard Edition databases
• Add HA to single instance databases.
• Realize millions of dollars in cost savings
See Whitepaper “EMC IT’s Virtual Oracle Deployment Framework”, November 2011
21
Confidential
vFabric Data Director
Data Director will help you:
1. Stretch Your Servers
2. Liberate Your Licenses
3. Double Your DBA Productivity
4. Secure Your Stuff
22
Confidential
How Many Databases Can a DBA Manage?
23
Confidential
Data Director Accelerates Productivity
After
Before
Self Service
Developers/Adminis
trators
DBAs
IT
Policy
Database
Team
Foundation
Tickets
24
Developers
IT
Confidential
Data Director Architecture
Data Director Application
DB Management Server
Linux
Database
Database
Database
Monitoring Agents
Monitoring Agents
Monitoring Agents
Operating System
Operating System
Operating System
vCenter
ESXi
25
Shell Virtual Machines
Database Template Library
Confidential
TEMP
T
SWAP
LOG
DATA
Create Your Own Databases, Virtually
Data Director Application
DB Management Server
Database
Monitoring Agents
Linux
Shell Virtual Machines
Database Template Library
TEMP
T
Operating System
New Database Based Linked to Parent Template
Database
Monitoring Agents
Operating System
26
Confidential
SWAP
LOG
DATA
A Data Director Database
Uses binaries from linked clone
disks on the parent template
Has a unique ID
and IP address
Database
Monitoring Agents
Operating System
Runs software and database
processes locally (in the VM)
27
Stores data locally
(inside the VM)
Confidential
Upgrading or Patching a Database
Data Director Application
DB Management Server
Oracle 11.1 Template
Oracle 11.1
Oracle 11.2
Monitoring Agents
Monitoring Agents
Linux
Linux
Linux
DB Based on Oracle 11.2
11.1 Template
Oracle 11.2
11.1
Monitoring Agents
Linux
28
Oracle 11.2 Template
Confidential
Cloning a Database
Data Director Application
DB Management Server
Oracle 11.1
Monitoring Agents
Linux
TEMP
SWAP
T
Linux
DB Based on Oracle 11.1 Template
29
Shell Virtual Machines
Oracle 11.1 Template
Empty
DBVM
Exact
Clone
of DB
Oracle 11.1
Oracle 11.1
Monitoring Agents
Monitoring Agents
Linux
Linux
Confidential
LOG
DATA
Complex Operations Made Simple
Creating or cloning a database
takes about 5 minutes.
30
Confidential
Administrators Are Still In Control
Standards and Best Practices
are automatically enforced.
31
Confidential
Users Still Get What They Need
Users can still configure some
options and parameters, guided
by policy and best practices.
32
Confidential
Why You Should Care
Using Data Director, complex database
operations are automated and simplified.
1.
2.
3.
4.
5.
6.
33
Database Creation
Database Cloning
Data Movement
Backup and Recovery
Load Balancing
Database Retirement
Developers and Analysts can
create their own databases
without an administrator.
Confidential
How Many Databases Can a DBA Manage with Data Director?
We break the direct relationship
between DBAs and the number of
databases under management.
34
Confidential
vFabric Data Director
Data Director will help you:
1. Stretch Your Servers
2. Liberate Your Licenses
3. Double Your DBA Productivity
4. Secure Your Stuff
35
Confidential
Database Security is Baked In
Concepts like database expiration are
built into the provisioning process.
36
Confidential
Network Security is Baked In
Network Security is built into the
provisioning process as well.
37
Confidential
Centralized DBaaS monitoring
All service level actions are closely
monitored, logged, and tracked.
38
Confidential
A Real World Use Case
A Production Finance Database has data quality
problems and an analyst needs to examine some
highly sensitive data.
This is a very difficult challenge for:
1. Security
2. Compliance
3. Database Operations
4. System Administration
39
Confidential
Linked Clones of a Database
Data Director Application
DB Management Server
Oracle 11.1
Monitoring Agents
Linux
Shell Virtual Machines
Oracle 11.1 Template
TEMP
SWAP
LOG
DATA
T
Linux
DB Based on Oracle 11.1 Template
Empty
DBVM
Linked
Clone
of DB
Oracle 11.1
Oracle 11.1
Monitoring Agents
Monitoring Agents
Linux
Linux
∆
The linked clone is a separate DB that
shares the original storage and keeps
a local copy of disk block changes.
40
Confidential
A Real World Use Case
Data Director can:
• Create a linked clone of the original database that
has an identical O/S and DB configuration.
• Ensure changes do not impact original DB.
• Automatically discover and mask sensitive data.
• Expire and delete the linked clone after 7 days.
• Automatically remove any backups or snapshots.
• Quickly create and apply a patch any databases
with a similar configuration.  Security



41
Confidential
Compliance
Database Operations
System Administration
Summary
vFabric Data Director enables database-aware virtualization on
VMware vSphere and provides database-as-a-service for
heterogeneous databases and Hadoop
Reduce cost through database-aware virtualization
Increase IT agility
Accelerate analytics and application development
42