1.What are the needed ingredients to successfully deliver on prem services
2.Why EMC ScaleIO is the best fit to deliver Elastic Block storage services
3.Real life architecture examples
1. Elastic Block
storage services on Prem
Antonio Romeo Presales Director & Eitan Segal Advisory Systems Engineer
EMC Software Defined Solutions - EMEA
3. Agenda
• Why “Elastic Storage” transformation:
Current Challenges
• The world of Structured Data
• Three steps to elastic block storage
4. Why we seek for Transformation
CEO
OBJECTIVES
Business
Opportunity
LINE OF BUSINESS
OBJECTIVES
Innovative
Business
Agile
IT
CIO
IMPERATIVE
MUST
HAVES
Increase
growth and
lower costs
Time-to-value
and operational
simplicity
Fast and new
business
models
5. • Reduction of storage CAPEX for the next 4 years
• Expected Growth Rate 20% YY
• Need to allow for faster, elastic provisioning of resources
• Provide operational efficiencies to reduce costs and provide agility to respond to the
business faster
• The use of Tier-1 or Tier-2 storage was based on location rather then using Block Service
SLA’s
• The actual storage and tier assignment was decided by the admins
• Day-1 operations ( provisioning ) was done by several teams each which his own toolset
• Production and test environment weren't logically and physically separated
• Block Storage SLA’s were defined but not linked to the type of applications or environment
• Lack of visibility created a big challenge to manage capacity and growth requirements.
Solution
Business Challenges
Current State
• In the next slides we will demonstrate the disruptive approach we took to transform this
customer to deliver Elastic Block Service
• It’s a very good example of “one size doesn’t fit all”
Enterprise use case
10. T-1 Production Apps
High performance OLTP
T-2 Production Apps
Non Production Apps
Dev. & test
Sandbox
Service
Definitions
Block Storage Capacity
Block Storage Performance
Block storage Life cycle
Align with specific OS
,virtualization etc..
T-Shirt
Definitions
Type of Protection Tiers
Local Protection
App Protection
OS protection
Type of Disaster Recovery
Active Passive
Application Replication
level
Backup Data Policy definitions
OS Data
Data
Retention
Data Protection
Definitions
Driving Standardization
11. What Ingredients we need
STANDARDIZED
Services
Policy based
Automation
Control &
visibility
New HW
Architecture
Software Defined
Technology
12. Software Defined
Data Center
• Service-based cloud
scale datacenter
• Auto Scale
• Hardware of Choice
Software Defined Storage
• Truly Elastic
• Multi Tenancy for service separation
• Agnostic to Type of Hypervisors, OS, Compute HW
• Supports blend of tiers, low-latency, capacity optimized
• Choice of architectures; Nodes, Rack-scale, Hyper scale-
out
• Automate – Open standard API support
• Enable Business to gain “cloud scale” economics within 3-
4 years TCO Windows
Focus on
Software
Defined
Technology
How SDS Helps us
13. New HW
Architectur
e
TO
Traditional
• Great Fit for dedicated
workloads and specific use
cases
• Scale Up architecture
• Can be used in a converge
model
Servers
STORAGE
FC
Network
SAN
SAN.NEXT
• Scale Out Storage on COTS
• No Storage Siloes
• No expensive FC
• Add storage independently to the App Hosts
ETH/I
B
APPS Servers
SDS Servers
Infrastructure .NEXT
ETH/IB
APPS &
STORAGE
A A A
A SApplications SDS
S S S
• Greater cost saving and efficiency
• Cloud scale and elasticity
• Infrastructure
mirrors IaaS consumption model
New Architecture
14. PRE-DEFINED Elastic
Block Services
Financial
Transparency
Operational
Management
Self-Service
and Automation
Traditional
Arrays
All
Flash
COTS
Hyper-
ConvergedConverged
VIRTUALIZED
Infrastructure
Storage Control Plane
SDDC
Architecture for Elastic Block
Storage Tiers
Automation and Service Level
Selection
ELASTIC BLOCK-SERVICE
Platinum
Gold
Silver
Protection Tier
Add Disks
Tier Policy Automate
Disaster Recovery
Self Service for Day-2 Operations
Expand Capacity Manage Snapshots
Env Selection; Prod
, Dev./Test
Move between Policies
Bronze
MONITOR SERVICE SLA SELF SERVICE COST TO SERVE
16. Next Gen Storage Infrastructure
73%
7%
20%
End State Storage Layout
ScaleIO
All Flash
Tier-1
SDS Layout
• ScaleIO SDS using Hyper-converge
Architecture
• Hyper converge Nodes in a VMware
environment and Linux Bare metal
• SIO Storage Nodes for mix storage
req and for 2 layer architecture
• 70% of the storage estate wont
need data migration due to EOL
HW-just add/remove servers.
• Most of Storage and Application
will run on the same server
* 4 year Transformation plan
17. What Elastic and Open really means
Scale Out
Balance Rebalance Rebuild
Open
• Support for any Hypervisor, OS
and HW Media
• Supports traditional and web
scale applications
• Different type of tiers
• Blend of type of nodes
• Auto balance, Rebalance and
rebuild
• Provides QOS per application
• Performance Scale Linearly
Elastic