Learn why Fujitsu is always the right way to do storage.

1. 0 Copyright 2017 FUJITSU
#FujitsuWT
Flash
Forward
Many new Ways
to Do Storage
D V Narasimha Rao

2. 1 Copyright 2017 FUJITSU
Digitalization increases data traffic
How do I find my new storage balance?
Response time
Capacity
Costs

3. 2 Copyright 2017 FUJITSU
Which way to go for finding a new storage
balance?
Hyper-Converged
SW-defined
Hybrid
All-Flash

4. 3 Copyright 2017 FUJITSU
Classical HCI
Starting Point – choosing the scalability
approach
Storage scales independently
NETWORK
COMPUTE
STORAGE
Integrated scalability
NETWORK
COMPUTE
STORAGE
NETWORK
COMPUTE
STORAGE
NETWORK
COMPUTE
STORAGE

5. 4 Copyright 2017 FUJITSU
Classical HCI
Part 1: Classical Scalability
Storage scales independently
NETWORK
COMPUTE
STORAGE
Integrated scalability
NETWORK
COMPUTE
STORAGE
NETWORK
COMPUTE
STORAGE
NETWORK
COMPUTE
STORAGE
HCI
Integrated scalability
NETWORK
COMPUTE
STORAGE
NETWORK
COMPUTE
STORAGE
NETWORK
COMPUTE
STORAGE

6. 5 Copyright 2017 FUJITSU
 Server virtualization, VDI
 Databases/OLTP
 Business applications
 Analytics (e.g. SAP HANA)
 Unstructured data
 Content depots/online archives
 Historical data for analytics
 Media streaming
Different Applications, different Requirements
Lowest costs of capacityReliable performance

7. 6 Copyright 2017 FUJITSU
Lowest costs of capacityReliable performance
Architectures for independent Storage
Scalability
SW-defined, hyperscale storage
 Extreme scalability
 Lowest costs (open
source)
 Object and OpenStack
Storage
Hybrid Disk und Flash Storage
 All-in one
 Balancing capacity, speed
and costs in one system
All-Flash Storage
 Fastest response time
 Lowest operational costs
(power,space,
maintenance, admin)

8. 7 Copyright 2017 FUJITSU
All-flash or Hybrid – Which Way to Go ?
Hybrid
All-Flash
 Price per capacity (flash/
disk) will reach break-even
within product lifecycle
 Data reduction can eliminate
price gaps short term
 Convincing savings in ops
and maintenance (TCO!)
When currently using mainly
fast disks  go all-flash
 Balance capacity, speed
and cost with SSDs
and Hi-capacity HDs
 Use automation tools to
reduce operational efforts
created by HDs
When consolidating structured
and unstructured data in one
storage  go hybrid

9. 8 Copyright 2017 FUJITSU
Software-defined Storage (SDS)
 High, fast and flexible scalability
 Extended lifecycle, fewer migrations
 Lower costs (open source SDS !)
Advantages
 Build your own storage
 Hidden costs – TCO risks
 Higher latency
Challenges
 Decoupling data management from HW
 Networked architecture
 Based on ×86 servers
Data Management Software

10. 9 Copyright 2017 FUJITSU
SDS needs the right-sized hardware
Look for complete
SDS solutions!

11. 10 Copyright 2017 FUJITSU
Current top usage areas of SDS
Part of hyper-converged
infrastructures
Object/S3
storage
Distributed
file storage

12. 11 Copyright 2017 FUJITSU
Demand for New Storage Class
Unstructured data, high
fraction of less frequently
used data
Operating system agnostic
interface
Global name space
Tolerate higher latencies
(10 – 50ms)
Hyperscale,
high bandwidth
No file locking
Versioning instead
Better segregation of storage
clients and data(ransomware
attack protection)

13. 12 Copyright 2017 FUJITSU
S3, The De-Facto Object Storage Standard
Firewall/Load-balancer
(http://s3.a.lan)
http Gateway 1
(http://s3-1.a.lan)
http Gateway n
(http://s3-n.a.lan)
…
Hyperscale
Storage
Operating system agnostic
Ransomware protection
by segregating clients & storage
Firewall crossing
Versioning instead of locking
Http based protocol
Global name space
multi-tenancy
Stateless, hyperscale storage

14. 13 Copyright 2017 FUJITSU
Why Open Source Based Object Storage?
Cloud native
applications are
the main driver for
data growth
Open to switch
vendor without
changing
technology
Hyperscale
storage innovation
is driven by cloud
native applications
More than 80%
of public cloud is
running on Linux
Open Source
software
supported by
communities
does not depend
on a single vendor

15. 14 Copyright 2017 FUJITSU
Classical HCI
Part 2: Hyper-converged Scalability
Storage scales independently
NETWORK
COMPUTE
STORAGE
Integrated scalability
NETWORK
COMPUTE
STORAGE
NETWORK
COMPUTE
STORAGE
NETWORK
COMPUTE
STORAGE
Classical
Storage scales independently
NETWORK
COMPUTE
STORAGE

16. 15 Copyright 2017 FUJITSU
Hyper-Converged Infrastructure Concept
VM VM VM VM VM VM VM
Administration
Network
Virtualization
Server
Virtualization
Storage
Virtualization
…
SSDs HDDs SSD HDDs SSD HDDs

17. 16 Copyright 2017 FUJITSU
Storage in Hyper-converged IT (HCI)
 Less fit for heterogeneous
storage workloads
 Maintaining quality of service for data
 Additional storage silo
Challenges
 Server in two roles:
compute & storage
 Virtual everything
(server, storage network)
 High and modular scalability
Advantages

18. 17 Copyright 2017 FUJITSU
HCI – Top Players
Compute and Storage Cluster
Storage Space Direct
Hyper-V Virtual Machines
Shared pool of local disks
Microsoft
vCenter
Cloud Foundation
NSX
vSAN
vSphere
Hardware Management Plug-Ins
SDDC Manager
Hardware Management
Services (HMS)
Pre-qualified vSAN Ready Nodes
(rack-mount servers incl. storage)
VMware

19. 18 Copyright 2017 FUJITSU
Hyperconverged = Simplification?
Look for integrated
systems!

20. 19 Copyright 2017 FUJITSU
Fujitsu: always the right way to do storage
ETERNUS AF
All-Flash
ETERNUS
CD10000
SW-defined
VMware
Microsoft
OpenStack
HCI
ETERNUS DX
Hybrid

21. 20 Copyright 2017 FUJITSU
Fujitsu
Storage
Data is our
Domain

22. 21 Copyright 2017 FUJITSU