The All-Flash Array for theNext Generation Data Center
Jeramiah DooleyPrincipal Flash Architect
@jdooley_clt
Now Part of NetApp
What Are We Talking About Today?
• Consumption Models
• Architecture Matters…But Why?
• What Happened to Storage?
• The Big Question
• Are We There Yet?
Modern Infrastructure Consumption Models
Best of Breed Appliance
As a Service
Hyper Converged
Converged Infrastructure
Software on Commodity Hardware
Now Part of NetApp
The Infrastructure Consumption Continuum
SMB / Startups Large Enterprise Hyperscale / SP
Best of Breed Appliance
Hyper ConvergedAs a Service
Software on Commodity Hardware
Converged Infrastructure
Ease of ImplementationEasier Harder
Degree of Vendor Lock-‐InMore Lock-‐In Less Lock-‐In
Implementation FlexibilityLess Flexible More Flexible
Cost EfficiencyAt Small Scale At Large Scale
SolidFire Consumption Models
Element Operating System
Converged Infrastructure
Agile Infrastructure
Best of Breed Appliance
SF Series Appliances
As a Service
Scale-‐Out Infrastructure Agility
GuaranteedQuality of Service
Complete System Automation
In-‐Line Data Reduction
Self HealingHigh Availability
Software on Commodity Hardware
element
Flash V2 - All Flash Array
Flash V1 PCI FlashFlash for the Next Generation Data Center
Requirements constantly changingConsumes the most capital and time
Applying All-‐Flash in your Data Center
PerformanceLow
Performance and Capacity(Public / Private Cloud, IaaS)
High
Cheap and Deep [object / file](Backup / Archive / Video Streaming)
Performance at any cost(Transactional Processing / Trading Platforms)
# of Applications
New Tech, Same Problems
Operational CostsData Center Costs
Efficiency
…we could address any size pool of storage the business needed?
What If…
High performance storage systems designed for large scale infrastructure
Industry Leading Quality of Service (QoS)
Scale-Out Architecture• 4 – 100 nodes• 35TB – 3.4PB Usable Capacity• 200k – 7.5M Controllable IOPS
Simple all inclusive pricing model Direct Tier 3 Support
for every customer
Industry-standard hardware• 10 GigE iSCSI, 16/8 Gb FC
• Replication (synchronous / asynchronous)• Integrated cloud backup• Snapshots & clones• Global efficiencies• 256-bit encryption-at-Rest• Comprehensive logging• Cloud based monitoring• Secure multi-tenancy (VLAN support, volume access groups)• Simultaneous multiprotocol support (FC / iSCSI) • Deep integrations: VMware, OpenStack, CloudStack
All Inclusive Feature Set
Platform Building Blocks
2405 • 5 to 10TB effective capacity• 50,000 guaranteed IOPS
4805 • 10 to 20TB effective capacity
• 50,000 guaranteed IOPS
9605 • 20 to 40TB effective capacity
• 50,000 guaranteed IOPS
9010 • 20 to 40TB effective capacity
• 75,000 guaranteed IOPSSolidFire’s Effective capacity calculation accounts for Helix data protection, system overhead and global efficiencies including compression, deduplication and thin provisioning. SolidFire customers typically achieve an effective capacity
range of 5x to 10x the raw capacity depending on application workloads.
Scale-‐Out Wins
Performance
Capacity
43.6 TB 250,000 IOPS
60.8 TB 350,000 IOPS
52.2 TB 300,000 IOPS
200,000 IOPS35 TB PERFORMANCE VIRTUALIZATION (IOPS)
CAPACITY VIRTUALIZATION (GB)CPU RESOURCES (CORES)
SYSTEM MEMORY (GB)
CAPACITY PERFORMANCE
CPU MEMORY
Building Blocks of ResourcesNODES
NON-DISRUPTIVE
GRANULAR
MIX NODES
SCALE OUT/IN
NO FORKLIFT UPGRADES
SOLIDFIRE CLUSTER 1 SOLIDFIRE CLUSTER 2 SOLIDFIRE CLUSTER 3
2405
Gen X
Element OS | S C A L E O U T
4805
CAPACITY PERFORMANCE
CPU MEMORY
CAPACITY PERFORMANCE
CPU MEMORY
How QoS Works
VOLUME
Per Volume QoS Settings
Burst 2,000Max 1,000Min 100
APP BEHAVIORBurst
Max
Min
MIXED WORKLOAD ENVIRONMENT
100 Min
200 Max
300 BurstVolume
3
1000 Min
1500 Max
2000 BurstVOLUME
2
100 Min
500 Max700 BurstVOLUME
1
Element OS | G U A R A N T E E D P E R F O R M A N C E
CLUSTER
Simplified Management
CUSTOM INTEGRATION INTERFACES PLUG-INS TOOLS
Self-service Portals
Homegrown Apps
User InterfaceActive IQ
PowerShellSDKs
VMware vCenterOpenStackCloudStack
Element OS | A U T O M A T E D M A N A G E M E N T
API
TIME YOUR BUSINESS
DESIGNInitial size (GB or IOPS)
INSTALLATION OPERATINGGB and IOPS required
Account Creation
SCALINGGB and IOPS needed
UPGRADEDESIGNInitial size (GB or IOPS)
Anticipated growth over lifecycle of array
IOPS required lifetime of product
Node size selection
RAID type, RAID grouping
Drive type (FC, NL, Flash)
Drive quantities and ratio
Cache requirements
Drive Sparing
Aggregate/CPG
MetaLUN and Bin file layout
INSTALLATIONBin File/RAID Layout
Aggregate/MetaLUN creation
Snapshot spare spacing
Vol0/quorum drive layout
Aggregate snapshot frequency
Snapshot growth limits
OPERATINGGB and IOPS required
Account Creation
Host to HBA assignment and tracking
Host initiator count limits
RAID type required
Primary Control Node assignment
IOPS limitation concerns
Snapshot growth policy
Thin Provisioning (or not)
Aggregate/CPG assignment
40% CPU limit considerations
SCALINGGB and IOPS needed
Drive Shelves required
Power/floor space restrictions
Control node limitations (CPU storage)
UPGRADEHBA compatibility list
CPU utilization on control nodes
Maintenance window
Mitigate risk during node outage
Internally Automated Storage System TasksOTHER VENDORS
Element OS | A U T O M A T E D M A N A G E M E N T
VDI Initiative
Dev/Test Initiative
Application
Start With One Project … Transform As You Go
NodesIT Projects Automation SOLIDFIRE
CAPACITY PERFORMANCE
vCenter
Power-Shell
SPBM
vRealize
VDI Initiative
Database
Cloud Initiative
Application
Self-Service IT
Fuel the Software-‐Defined Data Center with SolidFire
INTEGRATION
SOLIDFIRE
AUTOMATION
BUSINESS VALUE
Automated Management
GuaranteedPerformance
Global EfficiencyScale-‐Out
Data Assurance
vCenter Plug-in PowerShell Storage Policy Based Management (SPBM)
End UserComputing
Infrastructure Delivery
ApplicationDeployment
SIOC + QoS VAAI VASA
VM
SRM VVol
Virtualization Platform
Business Value
vRealizeSuite