Storage Array Technology

Security Level:
Contents

1 Storage Array Components & Architecture

2 Storage Array Technologies

3 Huawei Storage Products

1 Huawei Confidential
Components of Storage Arrays with Disk and Controller Integration

3.5-inch disk controller 2.5-inch disk controller

SAS expansion ports Interface modules

 Two SAS expansion  Two slots for hot-swappable interface modules
ports per controller  Types: 8 or 16 or 32 Gbit/s Fibre Channel, GE,
Power/BBU/Fan modules 10GE electrical, 10 Gbit/s FCoE, 10GE optical,
 1+1 Onboard ports 25GE optical, 40GE optical, 100GE optical, 56
 Up to 94% of power  5300 V5: four GE ports per controller Gbit/s IB, 10 Gbit/s FCoE, and 12 Gbit/s SAS
conversion efficiency  5500 V5: four SmartIO ports per controller interface modules
 AC, –48 V DC, and 240 V DC  Note: OceanStor 5300 V5 does not support 56
Gbit/s IB or 10 Gbit/s FCoE interface modules.

2 Huawei Confidential
Architecture of Storage Arrays with Disk and Controller Integration

Interface Interface Interface Interface

module A0 module A1 module B1 module B0 Service subsystem
8 x PCIe GEN3 8 x PCIe GEN3

PCIe Mirrored
Channel
Controller A Controller B

... Disk subsystem

Disk 0 Disk 1 Disk 23 Disk 24

12 V 12 V

Electromechanical
Power/BBU/Fan module 0 Power/BBU/Fan module 1 subsystem

Note: Service channel

The preceding figure shows the typical system architecture of Huawei converged
Management channel
storage with the disk and controller integration design, including OceanStor 2200
Power supply
V3, 2600 V3, 5300 V5, and 5500 V5.
3 Huawei Confidential
Components of Storage Arrays with Controller and Disk Separation
BBU modules
 2+1
 AC power failure protection

Controllers
 Dual controllers
 Automatic frequency adjustment for reduced power consumption
 Built-in fan modules (fan modules are integrated in controllers, but can
be maintained independently)

Management modules
 1+1
 Hot-swappable
 Multi-controller scale-out and interconnection for establishing
heartbeats

Power modules
 1+1
 Up to 94% of power conversion efficiency
 240 V DC

Interface modules
Note:  16 slots for hot-swappable interface modules
The preceding figures show the front and rear views of an engine  Types: 8 or 16 Gbit/s Fibre Channel, GE, 10GE electrical, 10 Gbit/s
containing two controllers. The front and rear views are almost the FCoE, 10GE optical, 12 Gbit/s SAS, 10 Gbit/s FCoE, and 56 Gbit/s IB
same for an engine containing four controllers. interface modules

4 Huawei Confidential
Architecture of Storage Arrays with Controller and Disk Separation
Interface module Interface module Interface module Interface module
A0 ...
8 x PCIe GEN3
A5 B5 ...
8 x PCIe GEN3
B0

Fan module 0 PCIe Mirrored Fan module 0

Channel
Fan module 1 Controller A Controller B Fan module 1
Fan module 2 Fan module 2 Service
subsystem
Interface module
C0 ... Interface module
C5
Interface module
D5 ... Interface module
D0
8 x PCIe GEN3 8 x PCIe GEN3

Fan module 0 PCIe Mirrored Fan module 0

Channel
Fan module 1 Controller C Controller D Fan module 1
Fan module 2 Fan module 2

Management
Management interface Management interface
module A module B subsystem

12 V
12 V
Power Power Electromechani
3 1 BBU BBU BBU BBU Power Power cal subsystem
3 2 1 0 2 0

Note:
The preceding figure shows the typical system architecture of Huawei Service channel
converged storage with the controller and disk separation design, including: Management channel
1. OceanStor 5600 V5/5800 V5 (dual controllers per engine) Power supply
2. OceanStor 6800 V5/18500 V5/18800 V5 (four controllers per engine)
5 Huawei Confidential
Storage Expansion Technology

Heavy load

Normal load

Light load

Initial Scale-up Scale-out Scale-deep Scale-in

Scale-up: When a storage system runs out of space, more disk enclosures are added for capacity expansion.
Scale-out: Once the performance and/or capacity limits of the storage controllers are reached, the only option is to add new
controllers to the storage system, and the workload is balanced between controllers.
Scale-deep: Scale-deep is firstly introduced by HDS VSP storage systems and can integrate 3rd party (external) storage systems
for virtualization (addition of tier‐2 and tier‐3 performance and capacity).
Scale-in: Scale-in is firstly introduced by Huawei HVS storage systems, which apply internal technologies to increase overall
storage system performance.

6 Huawei Confidential
SAS Cascading Technology for Scale-Up

 Traditionally, organizations world-wide have

Application servers utilized scale-up architecture when it comes
to storage. Traditional scale-up architecture
utilizes a dual-controller set-up, and adds
FC/iSCSI connection storage drives as needed to increase storage
capacity and performance.

Storage controllers  Scale-up architecture is used for small-

Disks and JBODS
• Provide basic storage medium enterprises because of cost effective
• More disks or
services for block/file usage.
enclosures will be and simple system for maintenance.
• Provide advanced software
added to the
storage system
features: HyperSnap,  When storage capacity comes to a bottleneck,
HyperReplication, QoS, etc. more disks or disk enclosures will be added.
when capacity is
Typically customers will deploy
increasing with The connection for disk enclosures usually is
the storage system from a small
the business scale. SAS cascading connection.
scale, for example, 2 controllers,
2 or 3 disk enclosures.

7 Huawei Confidential
SAS Cascading Connection Example
SAS Cascading Principles:

 Based on the build-in coffer disk or

external coffer disk mode, the SAS
cascading connection is different, to
make sure the storage controller can
access the coffer disk along the short
path.
 The SAS cable should be connected
as loops, to optimize the path length
for each enclosure, also to make sure
each enclosure can have redundant
paths.
 Typically, a maximum of eight
Build-in coffer disk enclosures can be connected to a
pair of SAS ports (depending on the
storage type and model), while two is
recommended.
External coffer disk
8 Huawei Confidential
Storage Controller Scale-Out

Controller enclosures are connected using IP switches. Each controller has a 2-port interface module that connects to two switches for redundancy.
The management network uses daisy chains to manage all of the controllers and IP switches.

9 Huawei Confidential
Scale-Up vs. Scale-Out

Item Scale-Up Scale-Out

Component Disks Controllers

Mode SAS cascading Controller interconnection

With the workload growth of

The disk capacity is insufficient, and the
Scenario applications, the controller
controller performance is not a bottleneck.
performance becomes a bottleneck.

1. Simple: Add disks to the system. 1. Flexible expansion based on

2. Cost-effective: costs on disks and requirements
related enclosures, no other expense 2. Consolidated storage data pool,
Benefit
3. Silo building: The maintenance and easy and simple to manage
future evolution will be complex and 3. Higher CAPEX (per TB) than
may require data migration. scale-up

10 Huawei Confidential
Standard Disk Enclosure

Disk module Disk module

Power module
Fan module

Expansion module Expansion module

Power module
2 U disk enclosure: 25 x 2.5-inch disks
4 U disk enclosure: 24 x 3.5-inch disks

Disk module

The disk enclosure and controller enclosure are integrated.

Power module The power button of the disk controller is invalid and does not
support power-on and power-off. The controller enclosure
delivers the power-on and power-off commands.
Expansion module

2 U 36-slot NVMe enclosure

11 Huawei Confidential
High-Density Disk Enclosure

Power module

System subrack

Fan module
Disk module

Expansion module

4 U high-density disk enclosure: 75 x 3.5-inch disks

12 Huawei Confidential
Contents

1 Storage Array Components & Architecture

2 Storage Array Technologies

3 Huawei Storage Products

13 Huawei Confidential
Pre-Copy Technology
The disk pre-copy technology enables the storage
system to monitor the disk status. When a member

﹢
1
disk in a RAID group is about to fail, data on the
Normal status
The system monitors disk is pre-copied to a hot spare disk, reducing data
the status of all disks.
loss risks.
Data disks Hot spare disk
The disk status is determined by reading the
attribute of the self-monitoring, analysis, and

﹢
2 reporting technology (S.M.A.R.T.). When the disk is
When one disk is about
to fail, the data will be about to fail, the pre-copy technology enables data
copies or backed up to on the disk to be copied to and backed up on the
a hot spare disk.
Suspected Hot spare disk
hot spare disk. By backing up data quickly to
faulty disk shorten or avoid long-time reconstruction after the
disk fails, the disk failure rate during reconstruction
is reduced and the product security is improved.

﹢
After the disk is
Huawei RAID 2.0+ technology uses hot spare
replaced with a new space instead of physical dedicated hot spare disks,
disk, the data will be
copied to the new disk. the copy/backup time will be reduced remarkably,
New disk Hot spare disk so that the multi-disk failure rate is reduced during
the pre-copy or copy-back period.

14 Huawei Confidential
HDD Bad Sector Repairing Technology
Disk bad sector repairing: The basic working mechanism is to mask the disk bad sector address, then map this
address to the disk's internal reserved space.

Automatic R/W failure analysis

Disk scanning technology
If disk read/write fails, the storage system will
comprehensively analyze the system status, disk status, Scan and check the disk medium during off-peak
I/O failure information, to determine whether this failure is hours to find out the bad sector, in case multiple bad
caused by disk bad sectors. sectors are accumulated.

Bad sector writing Bad sector reading

When writing data to the bad sector, the disk will When reading data from the bad sector,
automatically mask the bad sector address, then map the disk will recover the data using RAID,
this address to disk's internal reserved space, and the then write the data to the disk's reserved
data will redirect to the space. space.

15 Huawei Confidential
SSD Protection and Repairing Technology
Data Redundancy Technology Wear-Leveling
Rebuild of bad page using redundancy information
⊕ Erase Cycle Erase Cycle
100% 100%

50% 50%
x
Uncorrectable
Block Block
① NAND-level multiple copies & RAID: metadata (multiple copies), user
data (RAID) ① Wear-leveling: scheduling migration, the less wearing blocks will be
② Data recovery: LDPC, read retry, and intra-disk XOR with redundancy. reused.

Bad Sector Management & Background Monitoring Advanced Management

Logic Block Physical Block Data management module
Block 0 Block 0
Block 1 Invalid Block ②
Block 2 Block 2 Disk management module
Block 4 Block 3

②
Unused ①
Unused DISK
SSD DISK
SSD
①
Block 5 Block
Unused Reservoir

① Background monitoring: with read/write checking, bad sector reporting ① Online self-recovery: online factory reset of SSDs
② Bad sector isolation: bad sector monitoring, migration and isolation. ② DIE invalidation: self-reporting and reducing the capacity for use

16 Huawei Confidential
Coffer Disk Technology
 For Huawei OceanStor storage, coffer disks are used to
store three types of data: cache data requiring power
failure protection, OceanStor OS system data, and
system configuration information and logs. The storage
system has two kinds of coffer disks: built-in coffer disk
Coffer disks and external coffer disk.
 The first four disks in the controller (for build-in mode) or
in the first disk enclosure (for external mode) are coffer
disk. Coffer disks are configured with RAID 1, after
Build-in BBU storing the system data, other space is used for the
business data.
 Coffer disks can be used to protect cache data. The
BBU can provide power for the cache and coffer disks to
ensure that the cache data can be completely flushed to
the coffer disks.

17 Huawei Confidential
Hardware Redundancy Technology

 Storage devices are occasionally

affected by failures and disruptions
that are caused by network
Interface module connectivity issues, controller or
BBU module
general hardware failures. This is
called Single Point Of Failure (SPOF).
 A common approach to providing a
System Management module highly available service is to use
subrack redundant components. Redundancy
Power & Fan module
solutions are usually less expensive,
easier to implement, and easier to
manage.

Controller

18 Huawei Confidential
Hardware Redundancy Technology Details
CPU Redundancy
 At least two controllers in active-active or active passive mode are used.
 In active-active mode, the two controllers are both working and monitoring each other. If one controller fails, the other one will take over services
from the failed controller.
 In active-passive mode, one is working and the other is standby and monitors status. If the active controller fails, the standby controller will start
working and take over services from the failed one.

Power/Fan Redundancy
 Typically the power and fan modules are frequently failed because of working in high voltage and high-speed spinning.
 At least two power modules are configured and connect to the same circuit with load balancing. If one fails, the other one will increase the power
capacity for the entire system.
 Two or more fan modules work for one system. If one fails, the other one still works properly to dispatch system heat.

Battery Redundancy
 The write-back technology is usually used for better performance. With this technology, the writing acknowledge is sent back when data is just
written to storage cache.
 However, if the system is suddenly powered off, the (dirty) data will be lost because the cache cannot store data without power.
 Two or more batteries (backup battery units or BBUs) will be configured for the storage system. If the system is powered off, the batteries can flush
the cache data to disks for long-term storing.

Disk Redundancy
 Disks are prone to failures in the storage system, so Redundant Arrays of Independent Disks (RAID) is a basic and mandatory technology for all
storage.
 Based on the specific protection level and capacity requirement, RAID can be configured as RAID 0, RAID 1, RAID 10, RAID 5, RAID 6, or RAID TP.
 If a disk fails, RAID will reconstruct the data with XOR based on RAID level, then the failed disk can be replaced with a new one.

19 Huawei Confidential
Hardware Hot-Swapping Technology

 Some redundant hardware modules can be

replaced or added online (hot-plugging/hot-
swapping technology), including I/O
modules, fans, power modules, and disks.

Handle bar  Hot swapping is used whenever it is

desirable to repair a working system without
interrupting its operation. It may simply be
for convenience of avoiding the delay and
nuisance of shutting down and then
restarting complex equipment.

 Online expansion of I/O modules or disks

on the current system will also reduce the
costs in comparison to deploying a new
storage system.

20 Huawei Confidential
Power Failure Protection – Uninterrupted Power Supply

FC/IP Connection FC/IP Connection FC/IP Connection

Cache Cache
Cache
CPU CPU
CPU CPU
Back-end Bus Cache Battery
Back-end Bus
CPU CPU
Disk
Back-end Bus Build-in UPS
External UPS
 Unexpected power-off may cause the dirty data in the cache to lose, or even cause system hardware component failures
(such HDDs or fans in high-speed spinning).
 Uninterrupted power supplies (UPS) can provide power for system cache flushing or machinery component to cold down.
 Huawei OceanStor storage uses BBUs for power failure protection.

21 Huawei Confidential
Power Failure Protection – None-Volatile Memory

Application server
E
1. The application server 3. Data is locked and 1. Some storage systems use none-
initiates a write request written to the global
cache of node C.
volatile memory for cache protection
to node C.
in case of power failures.
2. Node C applies for a 2. The storage system will write the
stripe write lock from
the lock server. data to the none-volatile memory
CPU CPU CPU first, or back up the cache data in
the none-volatile memory. If the
Global cache
system is powered off unexpectedly,
4. The data is divided into fragments and saved to the NVDIMM.
the none-volatile memory still keeps
the cache data for flushing.
Storage Storage Storage
3. Huawei storage uses non-volatile
dual in-line memory module
Node A Node B Node C (NVDIMM), EMC storage uses non-
volatile memory (NVRAM).
Huawei OceanStor 9000 writing process

22 Huawei Confidential
Host Agent Technology
Host

HostAgent Application
1 2 Engine
 Host Agent is applied to flush the data in host
Activation Notification cache to storage systems, working with
snapshot, LUN copy, or replication features, to
6 3 Flash ensure the data consistency and availability
Host Status
Cache among the host, the storage system at the
Reset
production site, and the storage system at the
Host Cache
4 DR site.
1 2 3 4 5 6

7 8 9 A B C  Supported database applications: Oracle, DB2,

Sybase, SQL Server, and Exchange.
 Supported OS: Windows, Linux, and UNIX.
Execution
Disk 5 Completed  Host agent works with other software or storage
features to provide abundant BCDR solutions.
Snapshot

LUN copy
Replication

23 Huawei Confidential
Multipathing – Origination
Considering there is one LUN 0, there are two access paths to LUN 0, path 000 and path 100.
The following issues may occur:

X
0 0 0
Application abc 0 Application 0 abc
Application 0

0 0 Host/Server 0
Host/Server Host/Server
1 0 xyz
1 0 1 0

SPOF: If one path (for example, Confusion: The storage system Bandwidth waste: If the workload
path 000) is broken, the write recognizes two paths (path 000 is heavy, but all writes go through
request of the application to the and path 100), which have the path 000, the bandwidth of path
storage along path 000 will fail. same destination. Therefore, only 100 is idle and waste.
one data copy will be stored and
the other one will be discard.

24 Huawei Confidential
Multipathing – SPOF

Without multipathing With multipathing

Function Eliminating single
point of failure
Server Server
Multipathing
software

Application Scenario
HBA HBA

Multipathing technology can be

used for eliminating single point of
failures on HBAs, network
connections, controllers, or others to
LUN LUN LUN
LUN LUN LUN keep business continuous if any
hardware component failed.

Storage Storage

25 Huawei Confidential
 Multipathing – SPOF
Multipathing
Server Server Multipathing
software Server Server Working mechanism:
software

 The multipathing software will monitor the

I/O paths of HBAs, networking, and
controllers to detect the failed or
HBA HBA
HBA HBA
recovered path.
 Failover: When an I/O path failed, the
multipathing software will find out another
path on the current controller (if the HBA
or networking fails), or switch to another
LUN LUN LUN controller (if the controller fails).
LUN LUN LUN

 Failback: When the I/O path is recovered,

the multipathing software will switch the
I/O path to the primary one, then switch
Storage
Storage
services to the controller (if the controller
fails).
Failover Failback

26 Huawei Confidential
Multipathing – Load Balancing

Function
Without multipathing With multipathing Load balancing
Server Server

Multipathing
software
Application Scenario
Link
Multipath software can be
used for load balancing so
I/O Bandwidth
Bandwidth that the storage system can
expansion
bottleneck
utilize potential bandwidth,
then increase the storage
throughput.

LUN LUN LUN LUN LUN LUN

Storage Storage

27 Huawei Confidential
Multipathing – Load Balancing
Multipathing Multipathing
software Server software Server
Working mechanism:
 Typical load balance algorithms:
 Round-robin: Dispatches I/Os
to each path sequentially.
HBA HBA
 Least-I/O: Finds out the path
with the shortest I/O queue then
dispatches the I/Os to the path.
 Least-load: Counts the load of
each path and dispatches I/Os
LUN LUN LUN to the path with the lightest load.

Storage

Load balancing
28 Huawei Confidential
Mainstream Multipathing Software
Storage vendors generally use three types of multipathing software:
1. Self-developed software, such as EMC PowerPath, HDS HDLM, and Huawei UltraPath
2. Plug-in adapter based on OS multipathing, such as IBM and HP
3. Native OS multipathing, such as MPIO
Comparison: Native OS multipathing software is simple to use and can meet almost all requirements. Vendors' self-developed multipathing has better
reliability and easier customization for advanced features.

Windows Linux AIX VMware ESX Solaris

OS MPIO DM-Multipath MPIO NMP STMS

Huawei UltraPath UltraPath UltraPath PCM (based on MPIO) UltraPath UltraPath

EMC PowerPath PowerPath PowerPath PowerPath PowerPath

SDDDSM (based on MPIO) DM-Multipath SDDPCM (based on MPIO) NMP SDD

IBM
RDAC RDAC RDAC VERITAS DMP

SecurePath SecurePath SecurePath STMS

HP
DSM (based on MPIO) HP-DM HP-PCM (based on MPIO) NMP DMP

HDS HDLM HDLM HDLM HDLM HDLM

NetApp DSM (based on MPIO) DM-Multipath MPIO NMP STMS

Veritas DMP DMP DMP DMP DMP

29 Huawei Confidential
Thin-Provisioning Technology
Thin LUN Actual data Physical space 8 KB
32 MB
Pool
8 KB
32 MB
Space
allocation
8 KB 8 KB
32 MB
8 KB

Capacity
on write
1. Write
8 KB 8 KB
operation 8 KB 32 MB

2. Space allocation
or optimization Data volume: 32 KB Data volume: 32 KB
according to the mapping table Allocated space: 256 KB Allocated space: 128 MB

Technical principles Technical advantages

 Capacity on write: When the allocated space is insufficient, new  Efficient allocation policy: The 64 KB granularity
space is allocated with a 64 KB granularity by default. improves efficiency in allocating small data blocks.
 Mapping table: Logical units and physical units (with a default  Diversified reclamation mechanisms: supporting
64 KB granularity) are associated by a mapping table. reclamation by VMware VAAI commands, Symantec
 Redirection-on-time: redirecting I/Os from logical units to Storage Foundation commands, Windows Server 2012
physical units through a mapping table. commands, and full-zero page check reclamation

30 Huawei Confidential
Tiering Technology Overview

Multiple Single
Data Lifecycle
Applications Application

A single application with a

Multiple applications running Data lifecycle management
clear distinction between
on one storage system
hot and cold data

Meet performance and capacity requirements and reduce TCO

31 Huawei Confidential
Huawei SmartTier for Block

1. The storage administrator adds different types

Tiered storage pools of disks to storage pools and starts SmartTier.

2. SmartTier identifies I/O activities on each data

I/O monitoring block based on a data relocation unit.

3. Block Access Rate Ranking identifies hot data

Block Access Rate Ranking and cold data.

4. The data relocation module relocates data

Data relocation among storage tiers.

32 Huawei Confidential
Huawei SmartTier for File
NAS
Clients

NFS SMB File monitoring Collects statistics (file size/name/type/

a-time/c-time/m-time/uid/gid).

Pre-analyzes file capacity distribution

OceanStor according to the file attributes to help
File access rate ranking administrators understand space usage
Policy conditions:
and determine better tiering policies.
• Time-based
(creation, access,
modification)
SSDs HDDs Relocates files on SSD or HDD according
• File names File relocation to the policies, automatically or manually.
• File extensions
• File owner or group
• Quota tree
• File size
• SSD utilization

33 Huawei Confidential
SSD Cache Technology Overview

 Working principle
LUN/File System
 SSDs are used as the secondary cache for RAM cache.
 Read acceleration is available to LUNs and file systems.
Dedupe Meta  SSD cache partitioning is supported.
 Application scenario
RAM Cache  Random small-I/O read-intensive applications, such as
Oracle OLTP, SQL server, exchange, and file service
 Competitive advantages
SmartCache  Dynamic expansion and reduction, SSD resource quota
management, flexible open policy
 Read-only cache, SSDs without RAID protection, so
HDD SSD that capacity can be fully utilized

34 Huawei Confidential
Service Quality Assurance Technology Overview

Mixed applications with OLTP and backup Specific performance

varied importance levels co-exist requirements

Ensures performance of Restricts the maximum Allows users to set maximum or

mission-critical services in a performance of some services to minimum performance goals for
multi-service system prevent traffic burst from affecting the application
the performance of other services

Do you know exactly how to allocate storage resources to different services?

NO YES
QoS Cache Partition

Achieve specific performance goals for applications

35 Huawei Confidential
Huawei SmartQoS Technology
Medium-priority High-priority Low-priority
applications applications applications  Service Quality Control
– SmartQoS helps mission-critical services obtain storage
resources in a timely manner to realize their performance goals.

 SLA-based Management of Storage Resources

– Monitors application performance and achieves performance
goals.

 Policy-based Performance Optimization

Performance

– Sets performance goals for mission-critical applications.

– Sets performance upper limits for low-priority applications,
preventing resource contention.
– Sets QoS policies and implements them at different intervals.
Application

Enables the optimum storage resources to serve the most important applications

36 Huawei Confidential
Huawei SmartPartition Overview
SmartPartition partitions system cache resources and
protects the performance of mission-critical applications.

Host
concurrency Technical principles
 Divides system cache resources into cache partitions of
Cache
different sizes.
 Allocates cache partitions to different applications.
 Adjusts the host concurrency and disk access concurrency
capabilities for each cache partition, and controls the service
quality of each application.
Benefits
 Dedicated cache partitions for service isolation and high
reliability
Disk
concurrency  Reserved cache resources for mission-critical applications,
improving service quality
Service 1 Service 2 Service 3 Service N  Applicable when the storage system carries various services
at the same time

37 Huawei Confidential
Data Reduction Technology Overview

Saving OPEX Prolong SSD life

Saves capacity, power, maintenance, Inline compression and deduplication reduce the data
and OPEX from end to end written to SSDs, which prolongs SSD service life

Improving all-flash storage efficiency and extending lifespan

Some types of data cannot be reduced

The types include but are not limited to:
 Compressed data and databases, encrypted data
 Pictures, audio, and videos
 PDF and XML data

38 Huawei Confidential
Data Deduplication Mode
Post Processing Inline Deduplication Hardware Acceleration
(deduplication after storing) (deduplication before storing) (deduplication algorithm is implemented using
dedicated hardware, not using CPU resources)

Deduplication
hardware

39 Huawei Confidential
Snapshot Technology – Overview

Backup & Restoration Development & Testing Continuous data protection

Copy the
Activate a snapshot snapshot to HyperSnap
backup space Copy 1 Copy 2 Copy 3
Activate a snapshot
Fast rollback
Source Snapshot Backup
Source Snapshot

Rapid online data backup and restoration Development & Testing Repaired logical error

Zero backup and rollback window High space efficiency Zero data loss

40 Huawei Confidential
Snapshot Technology – Mechanism
Host write to D

Before snapshot After snapshot After data change

Source Snapshot Source Snapshot Source

Volume Mapping Table Volume Mapping Table Volume Copy-On-Write (COW)

A B C D A B C D A B C D1 D
Modified data
Old data

Data "D" is copied and relations in the mapping table are modified

Before snapshot After snapshot After data change

Source Source Source

Volume/ Volume/ Volume/
Active Snapshot Active Snapshot Active Redirect-On-Write (ROW)
File System File System File System

A B C D A B C D A B C D B D1 E1 E2
Deleted Modified New
data data data

41 Huawei Confidential
Continuous Data Protection Technology

Concept
 Continuous data protection (CDP), also called continuous CDP example
backup or real-time backup, refers to backup of data by automatically
saving a copy of every change made to that data, and allows the
user or administrator to restore data to any point in time.
LUN TP TP+1 TP+2 ... TP+N
 In an ideal case of continuous data protection, the RPO= 0, RTO ≈ 0.
 There are 2 types of CDP technology. True CDP-based solutions
can provide fine granularities of restorable objects, then it can back CDP CDP0 CDP1 CDP2 … CDPN
up and roll back data in any point of time. The near-CDP can only snapshot
Rollback Create
back up data in a specified interval. snapshot
duplicates

Application scenarios Readable and

writable snapshot
Snapshot
duplicate
Snapshot
duplicate
 Data protection in seconds: CDP snapshots restore data on the
source LUN quickly.
 Data analysis and testing: Duplicates can be created for the CDP
snapshots and mapped to hosts for data analysis and testing.

42 Huawei Confidential
Mirror Technology Overview

Protect data in heterogeneous storage High availability

1. Redundant data
backups for
heterogeneous
storage systems High availability
protection for
2. Accelerated local LUNs
performance of
heterogeneous
storage systems

Protect data in heterogeneous storage High availability due to doubled data

and improve performance copies

43 Huawei Confidential
Huawei HyperMirror Technology Mechanism
1. Create a mirrored LUN.
Converts a local or external LUN into a
Host mirrored LUN, creates a mirror copy, and
adds a mirror copy for the mirrored LUN for
redundancy.
2. Perform synchronization.
 Initial synchronization: Replicates all data from

Mirror LUN the mirrored LUN to the new mirror copy and
writes data concurrently to both mirror copies.
 Incremental synchronization: Replicates data
increments from the mirrored LUN to the
recovered mirror copy after it is recovered from
Splitting or Interruption.
A B 3. Implement mirror copy splitting.
The mirror copy is unavailable after splitting.
Mirror copy Mirror copy

44 Huawei Confidential
LUN Clone Technology Overview
A clone is created as an image of a LUN at the current point in time. Splitting a clone
creates a complete physical copy of the image.
HyperClone can be used for data analysis and testing.

④
②
①
③
Process Slave LUN
Master LUN Slave LUN Master LUN

1
Activate the HyperClone 2 synchronization is
Data 3 splitting, the
After 4 slave LUN can be
The
feature and start data complete. The master master LUN and slave mapped to host for data
synchronization. LUN and slave LUN LUN are independent analysis, testing, etc.
have the same data. of each other.

45 Huawei Confidential
LUN Copy Technology Overview
HyperCopy allows a target LUN to obtain
the full physical data copy of a source
Source Destination LUN at the current time. The target LUN
LUN LUN can be read and written immediately,
regardless of whether the data copy is
complete.

After a full copy is generated

Full Copy for the initial synchronization,
the follow-up incremental
Full copy is the copies will only duplicate the
complete data changed data from last copy.
duplication from the
source LUN. Incremental
Copy

46 Huawei Confidential
WORM Technology Overview
WORM: Write Once Read Many, write-once technology

Application Scenarios:
Financial securities (bills), government documents, legal
documents, and medical records.
Solution:
• Enterprise WORM (WORM-E)
• Regulatory Compliance WORM (WORM-C, SEC Rule
17a-4 & 17a-3)
WORM is widely adopted in archive solutions.
After Optical disc, tape library and disk storage (NAS) support
Write WORM.

47 Huawei Confidential
Multi-Tenant Technology Overview
Storage Multi-Tenant efficiently separates the resource
administrator data of tenants and assigns the management
work of some resources to tenants to make
tenants manage resources in their own virtual
storage spaces.

 Rights- and domain-based management

The storage administrator can allocate LUNs to tenants, and
assign and manage tenant administrators based on different
right privilege.
 Resource management
A storage administrator has full resource management
permissions. A tenant administrator can only manage resources
within its virtual storage space.
 Data protection
A tenant administrator can independently configure data
A storage administrator manages all resources in a storage protection functions for resources in its virtual storage space
array and all tenants.
and totally independent to other tenant.
A tenant administrator can only manage resources in its virtual
storage space.

48 Huawei Confidential
Storage Management Software

 Cloud Storage Management  Storage Networking Management

Massive Capacity Management and Backup, DR and iNVS, Providing Good SLA
Optimization
• Batch deployment and • DiskGuard management
management • DR management
• Resource monitoring, • Video surveillance
dispatching, forecast management
• Alarm management • Disk data collecting
• Privilege and domain Integrated Storage
…
management Solution
• Remote calling service

 Device Management
(DFCAPS Features)
• Device Discovery • User Authentication
• Fault management • Performance management
• Configuration management • Security management

49 Huawei Confidential
Huawei Storage Management Software
OceanStor
DeviceManager

For the management of

one device, including
discovering, configuration,
monitoring, etc.

eSight
OceanStor DJ
For the management of
For storage virtualization.
multiple devices, including
Based on an intelligent scheduling algorithm, DJ
networking, storage,
can automatically configure storage resources
servers, eLTE, UCC
and report device and service statuses.
terminals, etc.

50 Huawei Confidential
Storage Compatibility/Interoperability Overview
Storage is not a single system, it has to interact with other systems in
the data center, such as:
• Switches: FC / iSCSI
Application • Hosts: X86-64, Power
• HBA: Qlogic, eMulex
• Hypervisor & OS: VMware, Hyper-V, KVM, Linux, Windows, AIX
• Applications: Oracle, SAP
• Others: Backup software, storage virtualization
Hosts Hypervisor Operation System
Storage interoperability/compatibility provides seamless resource
consolidation and cooperation among a large number of heterogeneous
storage resources by using standard interfaces.

Switch

Huawei Storage Interoperability Navigator

IT Open Lab
Storage

51 Huawei Confidential
VMware Interoperability – VAAI

VMware vSphere® Storage APIs – Array Integration (VAAI) define a set of "storage primitives" that enable the ESXi host
to offload certain storage operations to the array, which reduces resource overhead on the ESXi hosts and can significantly
improve performance for storage-intensive operations, such as:

VAAI-Block:
• Hardware Assisted Locking/Atomic Test & Set
• Full Copy/XCOPY
• Write Same (ZERO)/Block Zero
• UNMAP
• ThinProvisioning

VAAI-NAS:
• Full File Clone
• Fast File Clone/Native Snapshot Support
• Extended Statistics
• Reserve Space

52 Huawei Confidential
VMware Interoperability – VASA
vStorage APIs for Storage Awareness (VASA) is a set of application program interfaces (APIs) that enables vSphere vCenter to
recognize the capabilities of storage arrays. Such capabilities, including RAID, native thin provisioning and deduplication, are
made visible within vCenter.

VASA providers communicate with Virtual Center to indicate storage topology, capability
and state information which supports policy-based management, operations
management and DRS functionality. VASA providers help to identify trends in a VM's
storage capacity usage for troubleshooting, correlate events on the datastore and LUNS
with a VM's performance characteristics, and monitor health of storage.

53 Huawei Confidential
VMware Interoperability – VVoL
VVoL (Virtual Volume) is a provisioning feature for vSphere 6 that changes how virtual machines (VMs) are stored and managed.
VVols transforms SAN/NAS into VM-centric devices by making the virtual disk a native representation of the VM at the array level,
and enables app-centric, policy-based automation of SAN and NAS based services.

Benefits of VVOL:
• Enables an administrator to apply a policy to a VM
which defines the various performance and service-
level agreement requirements, such as QoS, replication
or deduplication.
To use VVOLs, the storage hardware has to support the
vStorage APIs for Storage Awareness (VASA).

54 Huawei Confidential
VMware Interoperability – SRM
Site Recovery Manager (SRM) leverages VMware Infrastructure to deliver advanced disaster recovery management and
automation to ensure it is rapid, reliable, manageable, and affordable. VMware Site Recovery Manager integrates with third pa rty
storage arrays and replication appliances to provide a complete integrated Business Continuity solution. This integration is
achieved through a unique Storage Replication Adapter.

Function of SRM:
• Planned Migration
• Disaster Recovery

55 Huawei Confidential
Contents

1 Storage Array Components & Architecture

2 Storage Array Technologies

3 Huawei Storage Products

56 Huawei Confidential
Gartner Storage Level Definition – Storage Band

57 Huawei Confidential
Entry-Level/Middle Level/High-End Storage

Simple/affordable/smart Converged/Flexible/simple High reliability/performance/scalability

24/7 available Rich features to meet diversity Rich data optimization and protection features
Secure & online Ensures 24/7 service availability 24/7 availability & QoS
Start small & grow as required Universal storage in datacenter Mission-critical storage

OA/file/backup VDI/VM/DB Core DB/Core VM

Middle-level High-level

Entry-level

IDC Band 1 Band 2 Band 3 Band 4 Band 5 Band 6 Band 7 Band 8 Band 9 Band 10
0k-4.99k$ 5k-9.99k$ 10k-14.99k$ 15k-24.99k$ 25k-49.99k$ 50k-99.99k$ 100k-149.99k$ 150k-249.99k$ 250k-499.99k$ > 500k$

58 Huawei Confidential
Huawei Storage Product Portfolio & Positioning
Solutions

FusionData DR Backup Archive

Management

Device Management Intelligent Storage Management Intelligent O&M

Data

DeviceManager OceanStor DJ eService

 a All-Flash Storage Hybrid Flash Storage Distributed Storage Edge Data & Storage
Intelligent
Storage

OceanStor 100D FusionCube

OceanStor Dorado OceanStor Dorado OceanStor 18500/18800 V5 OceanStor 6800 V5
2000
18000 V6 8000 V6

OceanStor Dorado OceanStor Dorado OceanStor OceanStor OceanStor

5300/5500/5600/5800 V5 2200/2600 V3 OceanStor 9000 FusionCube
6000 V6 5000 V6 Dorado 3000 V6
1000

59 Huawei Confidential
Thank you. Bring digital to every person, home, and
organization for a fully connected,
intelligent world.

Copyright©2020 Huawei Technologies Co., Ltd.

All Rights Reserved.

The information in this document may contain predictive

statements including, without limitation, statements regarding
the future financial and operating results, future product
portfolio, new technology, etc. There are a number of factors that
could cause actual results and developments to differ materially
from those expressed or implied in the predictive statements.
Therefore, such information is provided for reference purpose
only and constitutes neither an offer nor an acceptance. Huawei
may change the information at any time without notice.