Upload
branden-armstrong
View
220
Download
1
Embed Size (px)
Citation preview
DS800-G25 Disk Array Pre-Sales Training
Storage Products Business Division
Agenda
2
1
2
Introduction of DS800-G25 Product
DS800-G25 Features
3DS800-G25 For Medium Data Center
Positioning
Middle and high-end storage products, high performance storage platform for medium data centers
Specifications
Dual controllers: 32GB or 64GB cache
Four controllers: 128GB cache, support for dual-write data
Architecture
Intel storage processor
Up to 516 disks
PCI-E2.0, SAS2.0 technology
1 / 10GE iSCSI, 8Gb FC host interface cards
Software Specification
RAID 2.0 technology, snapshot, local volume copy, remote volume mirror, thin provisioning, quad-controller cluster
DS800-G25 controller
DS800-G25 3U16 Exp
DS800-G25 2U25 Exp
4DS800-G25 Next-Generation Architecture
High-performance hardware technology
Controller Architecture Design
Modular Host Interface
The most high-speed bus bandwidth technology, single channel bandwidth 5Gb, 16 channel bandwidth 80Gb
5DS800-G25 Modular Redundancy Architecture
Modular redundancy design
— The main components such as power supply, battery, fan, and main control board are designed with redundancy.
A-A Controllers
— ALUA multipath mechanism
— Support path redundancy and load balance
Dual SAS link
— When a single link failure occurs, the path will be automatically switched
batteryfan Power supply
hard-disk cartridge
main control panel
interface card
Fan
main control panelhost interface card
controller machine framepower module
modular design
modular components
Cache Protection
— cache mirror, power failure protection
Agenda
6
1
2
DS800-G25 Product Introduction
DS800-G25 Features
Symmetrical dual active technology
Remote replication technology
RAID 2.0 and SSD cache technology
7DS800-G25 Active Disk Diagnostic
Error Repair
Disk Detection
DiskDiagnostic
Detect mechanism
Periodic safety monitoring for hard disks
When find the problem, according to the error code to determine error type
Start the repair mechanism according to the error type
Repair mechanism
Disk repair: rewrite the error block to repair
Repair bad block: By remap and RAID mechanism to replace bad block
RAID rebuild: rebuild data by optimized techniques RAID 2.0
Diagnostic function
Double disk detection: double detect to confirm whether the hard disk failure is resolved
Active prevention Quick Fix
Double Diagnosis
Target: To reduce 80% downtime caused by disk failure!
8 DS800-G25 RAID 2.0 Technology
Hard disk failure is a problem in the whole storage system.
Universal way to deal with in the industry: the RAID, but each RAID has significant shortcomings, how to
solve?
hard disk 1 hard disk 2 hard disk 3 Hot Spare
RAID Type RAID5 RAID6 RAID10
Advantage Less disk wastedAllows two drive failures
simultaneouslyUp to half disks failure
simultaneously
Disadvantage More than one disk failure will lose data
Waste more than 2 disks Waste half disks
RAID 2.0 technology
…… …… …… ……
No DataNo Data
DataData
UnitlUnitl
RAID 2.0 allows multiple disks medium error at the same timeDisk rebuild time is reduced by 80%
Customer value :
9DS800-G25 RAID 2.0 Technology
Storage Pooland cells
RAID Resources
LUN-Virtualization
High-speed SAS SSD Low-speed SAS
LUN0 LUN1
Tie0:SSD storage Tie1: High-speed SAS
Tie2: Low-speed SAS
RAID10
Tie0:SSD storage Tie2: : Low-speed SAS
RAID5 RAID10 RAID5RAID6
Disk Disk
Disk
Disk Disk
Disk Disk
Disk Disk
Disk
Disk Disk
Disk
Disk
Disk
Disk
Disk…
Disk Disk
Low-speed SAS
The RAID resource is divided into 1GB sub blocks, which are called unit
Different units in accordance with the application of the performance and capacity needs to be combined into LUN
Physical storage resources
Administrator Server
SSD
Disk
Disk
Disk
…
Server
Storage Pool 0 Storage Pool 1Storage Pool is comprised by units
RAID 2.0 unit : also called "cell", which is the basic unit of data and storage resource management
10DS800-G25 RAID 2.0 Technology
Storage Pool
RAID resources
LUN-Virtualization
High-speed SAS
LUN0
Tie0:SSD storage
Tie1 : High-speed SAS
Tie2: Low-speed SAS
RAID10 RAID5 RAID6
Disk Disk
Disk
Disk Disk
Disk Disk
Disk Disk
Disk
Low-speed SAS
The RAID resources are divided into sub-blocks 1GB
Physical storage resources
Administrator Server
SSD
Disk
Disk
Disk
…
Storage Pool 0
A single RAID group allows 3 disk failure (triple parity)
Greatly improve the speed of disk rebuilt, reduce the risk of data loss
Accurate data bad block early warning and handling mechanisms to reduce the risk of data loss
RAID 2.0 technologyRAID 2.0
technology
11DS800-G25 SSD Cache Technology
• The performance of the product is greatly improved
by the SSD hard drive.
– The random access performance of the product can
be promoted by the SSD hard disk.
– Increase MicrosoftSQLServer operating speed 3
times.
– Increase OracleOLTP operating speed 3 times
– Start 500 virtual desktops within 8 minutes.
– The time to manage and tune Microsoft SQL Server
is reduced by hours per week.
12DS800-G25 SSD Cache Advantage
SSD Cache improve the performance by 25
times
Before enabled: IOPS less than 1,500
After enabled: IOPS over 40,000
Second level cache performance improvement instance
First level cache performance improvement instance
The first level cache improve the performance
by 40 times
1GB cache: IOPS less than 3,800
20GB cache: IOPS over 150,000
1 TB Cache: 1GB 1 TB Cache: 20GB
13DS800-G25 VMware Performance Acceleration
VMware ESXi & ESXVirtual SMP
VAAI features support: full copy, block
zeroing, scalable lock management
To accelerate the clone, migration,
initialization of the virtual machine at the
storage layer
Enable hardware acceleration features,
performance can be improved by 15
times
link operation hardware acceleration
Result(Second)
Bandwidth (MB / s)
Gigabit
iSCSI
CloningOpen 586 87
Close 8871 5.8
migration
Open 649 79
Close 7436 7
8GbFC
CloningOpen 120 427
Close 328 156
migration
Open 110 465
Close 332 155
14High availability solution: dual active array
Dual active array solution: Use two storage arrays, and the data is
mirrored, if one of the storage array fails, then switch to another storage array
online. The RPO and RTO is 0 approximately.
15High Availability: Dual Active Array
application system Kinds of application services
Solution description:
Two DS800-G25 storage with
remote mirror features, to achieve the
data mirroring.
Server can simultaneously see the
main storage and mirrored storage.
When the primary storage is down,
it automatically switches to the
mirror storage.
Advantage :Based on the storage without third-
party software and hardware
The solution is simple, easy
Applications: high availability for critical business
Primary Storage
Production volume 1
Production volume 2 copies
Target storage
mirror
GE/10GE
16Difference between dual active and remote mirror of DS800-G25
Sugon dual active storage: automatic failover
Primary Storage
Production volume 1
Production volume 2 copies
mirror storage
mirror
GE/10GE
parimary path backup path
Primary Storage
Production volume 1
Production volume 2 copies
mirror storage
mirror
GE/10GE
parimary path
general way: manual failover
1. Mirror volume is not visible, it must be manually operated when to access2. Server and mirror storage do not have access path, you must manually switch
server server
1. Mirror volume can be seen, but can’t be written, when the primary storage fails, it automatically provides access 2. Server and mirror storage has pathes, when the primary storage fails, it can be switched in real time.
heartbeat
unified management
17Symmetric dual active solution
Production volume
mirrored volume
mirror
10GE
Host A Host B
Storage Engine A Storage Engine B
Virtual volume
Data Center
Production volume
mirrored volume
mirror
10GE
Host A Host B
Storage Engine A Storage Engine B
Virtual volume
Data Center A Data Center B
Dual active storage Dual active data center
RPO=0
RTO=0
Without any third party software and hardware
synchronic distance
18DS800-G25 dual active networking
Data Center
Host A Host B
FC/iSCSI
Production volume
mirrored volume
mirror
10GE
SP1SP1
SP2SP2
SP1SP1
SP2SP2
Storage Engine AStorage
Engine A
Storage Engine BStorage
Engine B
10GE10GE
PCI-EPCI-E
Production volume mirrored volume
Private networkPrivate network
Storage Engine A Storage Engine B
Networking description:
( 1 ) The connection between the storage and the server can choose
iSCSI, FC network.
( 2 ) The link between the dual storages is 10GE, each controller is
configured with one dual port10GE IO interface card, 4 direct optical fiber
with mesh connection.
( 3 ) The suggested link length (synchronization distance) is no more
than 300M.
( 4 ) Mainly used for the interconnection within one data center, or
dual active storage solution.
A schematic diagram of the mirror connection between the controllers
19Read and Write IO
Data Center
Host A Host B
FC/iSCSI
Production volume
mirrored volume
Storage Engine A Storage Engine B
Path A Path B
读
Data Center
Host A Host B
FC/iSCSI
Production volume
mirrored volume
Storage Engine A Storage Engine B
Path A Path B
Production volumes, mirror volumes can be received at the same time to write IO, the mirror volume of the IO request will be transmitted through the mirror channel to the main storage engine, complete the write operation.
Production volumes, mirror volumes can be read at the same time in response to read and write IO.
Write Process Read Process
Read IO Read IOWrite IO
Write IO
Write IO
mirror imageIO
20Symmetric dual active failover process
Primary engine failover process:Step 1: detect the failure of the storage engine A, start the storage switching.Step 2: mirrored volume is automatically promoted to production volumes.Step 3: the multipath software switch the path B to primary. Step 4: the application on the server will access the volume on the storage engine B.
When the primary engine fails, the business can failover in real time without affecting applications.
Data Center
Host A Host B
FC/iSCSI
Production volume
mirrored volume
mirror image
10GE
Storage Engine A Storage Engine B
Path A Path B
ⅹⅹ
ⅹ
20
21Symmetric dual active + replication / snapshot solution
Data Center B
The two storages configured with symmetric dual active function, can also be enabled with replication/snapshot function, to achieve comprehensive data protection.Symmetric dual active: Storage controller level protectionReplication / snapshot: Lun level protection
Data Center A
Host A Host B
FC/iSCSI
Production volume
mirrored volume
mirror
10GE
Storage Engine A Storage Engine B
Copying volume
Replication
10GE/GE
Disk array C
22Comprehensive protection of data
44% hardware failures
49% software /human / virus fault
Data center and disaster recovery center storage and data redundancy
软件故障,32%
硬件故障,44%
人为错误,14%
病毒影响,3%
自然灾难,7%
Disaster Types
7% natural disaster Business disaster recovery center to take over and restore
Data protection response
Data center or disaster recovery center for data protection CDP
Natural disasters
Virus effects
Human mistakes hardware
failure
software failure
23Data protection comprehensive implementation
LAN
Production volume copying volume
WAN
Production volume
copying volume
X
Hardware failure countermeasures -- Storage and data redundancy
Data Center
Data Center disaster recovery center
X
快照
快照
快照
生产卷A
Logical failure countermeasures -- CDP continuous data protection
Data center or shared disaster recovery center
Up to 512 images are protected
disaster recovery centerData Center
Natural disaster countermeasures -- Take over the disaster recovery center business
X
24Remote replication technology
LAN/WAN
Host
Production volume
Copying volume
New data block
On-demand select data
synchronization strategy
policy name triggering condition data difference Link requirements
Timed replication Fixed times, such as daily 12:00 hour low
Periodic replication Fixed period, such as every 30 minutes minute lower
Continuous replication
Single write IO trigger, real-time replication
IO higher
25Snapshot technology
9am
10am
11am
Automatic continuous data snapshot
–create the time point record automatically
for the data volume according to the
strategy
–Quick recovery in a few seconds when a
data loss or error occurs
512 Data Images
–Each volume can have 512 time point
record.
–Based on the incremental time point
record, 1.2-2 space is allocated.
Snapshot resource pool dynamic
extension
–The snapshot resource pool is
automatically increased, and the strategy
and location can be set flexibly.
147
10
258
11
369
12
Write new data block
10:00-10:59
147
10
258
11
369
12
9:00-9:59
147
10
258
11
369
12
Write new data block
26
7
910
2 6 7
9 102
67
Source Resources Automatically creates a snapshot
Snapshot resource
Data Storage Block
The initial(not containing any data).
001 002 003
…255
26Synchronous mirror technology
LAN/WAN
Host
Production volume Repllication volume
Synchronous mirror between two disk array
– Data is synchronously written to the primary storage
and disaster recovery storage, the mirrored volume is
one accurate and complete copy of the primary
volume
Data blocks are written to the main memory
14
Primary storage for the host returns successfully written confirmation signal
2 Data blocks are written to the backup copy volume
3 Return to the main storage in order to confirm the signal after writing success
Applications
– In close proximity or within the same city (low-
latency links), remote and recovery disaster recovery
storage zero data loss
27One to one disaster recovery solution
One-to-one disaster recovery modeHigh / low product collocation, lower cost
Support IPV4, IPV6
Flexible data synchronization strategy
Close sites, can be upgraded to dual active data center
Primary site Disaster recovery site
Mirror/Replication
policy name triggering condition data difference Link requirements
Timed replication Fixed times, such as daily 12:00 hour low
Periodic replication Fixed period, such as every 30 minutes minute lower
Continuous replication Single write IO trigger, real-time replication IO higher
On-demand selection of data synchronization strategy
28Disaster recovery solution of three centers in two cities
Production center Disaster recovery in the same city
Disaster recovery in different cities
Disaster recovery in the same city : Provide
application level data protection and recovery
capability
Disaster recovery in different cities : Provide
data level backup and recovery capability
A : Production
B : in the same city
C : in different cities
Mirror
Replication
No third party software and hardware, the storage array level to
achieve the protection of three centers in two cities
Close to disaster recovery in the same city, can be upgraded to
double active data center
Three networking modesThe best mode for disaster recovery
A : Production B : in the same city C : in different cities
Mirror Replication
BHOP
( Mirror+Replication)
1:2( Mirror + Replication )
A : Production B : in the same city C : in different cities
Replication Replication
BHOP
( Replication)
synchronization asynchronization
synchronization asynchronization
synchronization
asynchronization
29many to one disaster recovery / shared disaster recovery solution
Best practice
The Chinese Academy of Sciences Center for cloud
disaster recovery network : Across the 5 provinces of
Chinese, Chinese largest shared disaster recovery cloud
platform
Hubei finance shared disaster recovery : From 17 cities
to the disaster recovery center, Hubei's largest shared disaster
recovery platform
Distributed data center, centralized disaster recovery
64: 1 DR : To support the disaster recovery capabilities of
the 64 branch node on a central node
A variety of disaster grade : A branch node to reach the
standard grade 3-5 of the disaster recovery strategy, nodes do
not interfere with each other
Centralized and unified management : Through the IDSM
to achieve a unified disaster recovery management
WANcenter site
Site 1 Site 2 Site N
…
many to one disaster recovery / shared disaster recovery
Horizontal construction : Ministry / provincial / municipal
governments share a unified disaster recovery
Vertical construction : Focus on disaster recovery large
enterprises with many branches of the structure
30 Simple and easy to use - performance monitoring and alarm mode
— Read and write IOPS
— Read and write bandwidth
— Read and write response time
— CPU, storage usage
SMA Mail Buzzer Console Log
Data bad block ü ü
Media error ü ü ü ü
Hard disk failure ü ü ü ü ü
Fan fault ü ü ü ü
RAID fault ü ü ü ü ü
LUN failure ü ü ü ü ü
FC Target ü ü ü ü
iSCSI Target ü ü ü ü
Controller failure ü ü ü ü ü
Performance monitoring content:
— Indicator light alarm
— Console alarm
— Buzzer alarm
— Mail alarm
— SMS alarm
……
Alarm mode:
31 Simple and easy to use - management interface
Graphical User interface
— Administrators can easily grasp, simple and
easy to use
step - by -step configuration
— Just click the mouse, you can complete the relevant configuration
centralized manegement — On the same management interface to manage multiple devices
Sugon Building, No.36 Zhongguancun Software Prak,
No.8 Dongbeiwang West Road, Haidian District, Beijing 100194
Web : Http://www.sugon.com
THANKS