This document introduces H3C's IRF2 technology and products. IRF2 allows multiple physical devices to be virtualized as a single logical device for high performance, reliability, and manageability. Key benefits of IRF2 include doubled performance through active-active operation, centralized management through a single configuration file, and fast recovery time through elimination of STP. The document discusses various H3C switch products that support IRF2 including the S12500 series, which can virtualize up to 32 physical chassis into one logical device.
2. 1. Current Network Challenges
2. H3C IRF2 Introduction
3. H3C IRF2 Product Introduction
4. H3C Green IT
5. H3C IRF2 Case Study
2
3. The Challenge of IT Network
• Device Increase !!
• Complex Topology !!
• Difficult to Manage !!
• Compatibility !!
• High Reliability !!
• Load Balance !!
• Able to Failover ??
• Able / Easy to Recover
after Failover ??
3
4. Common Difficulties of Network Administrator
farm
erver
tion s
WhileApplica are more applications
there How to implement remote
And higher performance requirements, disaster recovery of data?
can the data center support future
service development?
How to guarantee security of key
What will be included in large- and data that faces soft and hard faults?
mid-scale networks?
How to construct them?
How to ensure network reliability
and service continuity?
How to manage the network in a
unified manner and reduce the
management and maintenance cost?
How to address the compatibility problem
of heterogeneous devices purchased
at different times?
4
5. Network Design Concerns
Challenge
VRRP configuration
Scalability & Availability has becoming more and more
important for campus network and data center
construction. IGP configuration
Conventional technology facing challenge
including : 1. Slow network convergence MSTP+VRRP design of
2. Complexity for management flow load balance and
VLAN assign
3. Low utilization of bandwidth
4. High cost for upgrade network
VRRP Multiple Instance MSTP Multiple service
Core
L3 NS
L2 FW
IPS
LB
VRRP instance Master MSTP instance
VRRP Multi-instance MSTP Multiple service
5
6. 1. Current Network Challenges
2. H3C IRF2 Introduction
3. H3C IRF2 Product Introduction
4. H3C Green IT
5. H3C IRF2 Case Study
6
7. What’s IRF2
Virtualization
H3C Compatibility
H3C’s Virtual LAN
Technology.
IRF2 For H3C Full Series Box
and Chassis Devices
Software
Application
Security
Network
Architecture
Expansibility
Create Advanced hardware
Framework & Integrate Multiple
Architecture Service and Openness
7
9. How we look at IRF2 Technically?
CPU
Master CPU
Engine
CPU
Slave
Engine CPU
CPU
Slave
Engine
9
10. Advantage of IRF2 – Doubled Performance
Active device
Active Link
Standby device
Traditional Network Standby link Vituralize by H3C IRF2
Technology
Without IRF, all core and distribution With IRF, all core and distribution
devices are working as active and are working both active---two
standby---only one at the same time
devices as one logical device
Without IRF, all links are also active
With IRF, all uplinks are working in
and standby---Only one link used at the
same time LACP mode---all load sharing
10
11. Advantage of IRF2 – Manageable
10.2.1.1
10.1.1.2 10.2.1.2
10.3.1.1
10.3.1.5
10.3.1.2
10.3.1.3
• Aggregated devices are treated as one logical device, all members are managed
by master device (or master MCU in chassis switch).
• Only one configuration file works on the entire stack.
• The entire stack is able to be managed via any port on its member.
11
12. Advantage of IRF2 – Resilience
Convergence
Time
Speed Network
WAN
IP=10.153.108.2/24
Convergence
Virtual IP=10.153.108.1
N Second IP=10.153.108.3/24
Gateway=10.153.108.1
Gateway=10.153.108.1 WAN
IP=10.153.108.1/24
IP=10.153.108.1/24 WAN
Second Gateway=10.153.108.1
IRF2
Gateway=10.153.108.1
IP=10.153.108.1/24
Gateway=10.153.108.1
ms
Gateway=10.153.108.1
VRRP MSTP IRF2 Deployment
12
13. IRF2 Benefits
H3C Intelligent Resilient Framework
2 (IRF2) is a network virtualization
Enhanced
technology that virtualizes several
Performance Easy
physical switches into one logic
Management
switch. In a IRF2 system, the data
plane and switch fabric of all
IRF2 Fabric engines are active at the same time,
thereby providing a multiplied
+ =
Virtual switching capacity.
Switching IRF2 members are connected by
virtual switch channels (VSCs).
VSCs use standard 10 Gigabit
Ethernet connections (as many as
twelve 10G trunk) between the IRF2
Multi-service members. VSCs can carry regular
user traffic in addition to the control
plane communication between the
High IRF2 members without
Reliability compromising functionalities.
Virtual Switching Easy Management Multi-service
Virtual L2 switching Centralized Management Sec-blade firewall
Virtual L3 switching Unified OS/CLI Load-balancing module
Virtual MPLS switching Netstream
IPS module
Enhanced Performance Wireless controller
High Reliability
Up to 10 units in an IRF2 Fabric 1:N backup
Multiplied performance Distributed link aggregation
Eliminating the dependency on Spanning Tree Protocol (STP)
13
14. IRF2 Benefits
IRF2 can run on H3C S5800, S5820, S5500-EI, S5120EI
box switches and S7500E, S9500E, S12500 chassis
switches. IRF2 enables the same series switches work as
a single logical unit in the network.
By applying IRF2, either chassis
switches or box switches can be
S12500 Series
virtualized as a single high reliable,
high throughput and high integrated Chassis Switch
logical device.
S9500E Series
Series
Chassis Switch
S7500E Series
Chassis Switch 5 Breakthroughs of IRF2
S5000
Series
Box Switch
Break through the limitation of regions
Break through restriction on performance & bandwidth
IRF2 Break through the constraints of slots and interfaces
Break through the restriction on the form of equipments
Break through the restrictions on services
14
15. IRF2 in SMB network
1500
User
100 200
user
user
Focus on
• Lower TCO
• Higher Price Performance
Ratio
• Pay while grow
15
16. Revolution of Campus LAN Architecture
H3C IRF2 is transforming enterprise network
architecture for simplyness and optimization
Common Network Stacking Network IRF2 Network
16
17. IRF2-based 4 Chassis Virtualization
4 x Chassis Switch IRF2
4 x Core switch IRF2
2 x Chassis Switch IRF2
2 x Core Switch IRF2
8 x Switch IRF
2003 2009 2011
17
18. H3C Optimizes data center with Intelligent
Resilient Framework 2 (IRF2)
L3
Fast Failover L2
MiliSecond
Recovery
MSTP instance
H3C IRF2 No STP, only link aggregation, miliseconds failover
10.1.1.2 10.2.1.2
10.2.1.1
Simplified
Management 10.3.1.1
Single IP 10.3.1.2
10.3.1.5
Management 10.3.1.3
H3C IRF2 With IRF2 stacking, multiple devices-single config file
Doubled
performance
Active-Active
mode
Two-tier network design suitable for
flat server-to-server connectivity H3C IRF2 All paths active, performance is doubled 18
19. H3C Chassis Base Switch
Double Fault Tolerance Network
10G
Any two link failure
Any two switch failure Whole network B&F transaction and data
recovery within base will not be affected
Any two switch milliseconds level
components 19
20. H3C “4-in-1” Intelligent Resilient Framework 2
(IRF2) virtualization solution
Dark Fiber /
DWDM
SITE 1 SITE 2
Dark Fiber /
DWDM Dark Fiber /
DWDM
H3C CLOUD NETWORK
SITE 4 Dark Fiber /
SITE 3
DWDM
H3C 4 CHASSIS VIRTUALIZATION
Four core switches dispersed
among multiple sites into a single
IRF2 logical unit 20
21. IRF2 in Enterprise network
Extranet
Intranet
Internet
OA Network
Core Area
OA Network
Data Center
WE
B
WE
B
WE
B
AP
P
AP
P
AP
P
DB DB DB IP
SAN
IP
SAN
IP
SAN
Focus on
• Architecture Flexibility
Management Software
• Multiple Services
• Performance &
Management Management
Server Server
Scalability
Management Area
21
22. IRF2 in Enterprise network
Focus on
• Architecture Flexibility
• Multiple Services
• Performance &
Scalability
22
23. H3C IMC total network management Solution
H3C IMC is designed exclusively to control a complicated, multi-
vendor, highly virtualized network platform across multi-sites. Its
unified management idea and single pane of glass make it unique in
the market.
23
24. 1. Current Network Challenges
2. H3C IRF2 Introduction
3. H3C IRF2 Product Introduction
4. H3C Green IT
5. H3C IRF2 Case Study
24
25. H3C Switch Product Portfolio
IRF2 IPv6 IRF2 IPv6 100G IRF2 IPv6 IRF2 IPv6 100G
Backbone
switch
series
S7500E S10500 S9500E S12500
M-
IRF2 IPv6 IRF2 IPv6 core IRF2 IPv6
GE switch
series
S5120SI/EI S5500SI/EI S5800
IRF2 IPv6
FE switch
series
S3100v2EI S3600v2SI/EI S3610
25
26. S12500 Product Introduction
100G Platform 4/8/32-port 10GE line card
Maximum 512 10GE ports 48-port GE
200ms buffer capability
S12508 S12518
Switching
3.06Tbps 6.66Tbps
capacity
Throughput 960Mpps/ 1920Mpps 2160Mpps/ 4320Mpps
Bandwidth 360Gbps/ 720Gbps 360Gbps/ 720Gbps
CPU 1+1 1+1
Switching fabric 8+1 8+1
CPU slots 2 2
Switching fabric
9 9
slots
I/O slots 8 18
GE (max) 384 864
10GE (max) 256 576
256Mbytes/10G port 256Mbytes/10G port
Buffer per port
21.3Mbytes/GE port 21.3Mbytes/GE port
Fan module Fan N+M, fan frame1+1 Fan N+M, fan frame1+1
Power module N+M N+M
26
27. S12508 Product Physical Appearance
Power cooling fan
Power Power cables
module
6 (max)
Fan frame
22 RU ( 975mm)
CPU engine
1 ~ 2 pieces
Switching fabric
I/O slots 9 (max)
8 (max)
Fan frame
442mm
27
28. S12500 Benefits
512MB 512MB CLOS multi-layered switching
Fabric architecture
512MB 512MB
Dynamic non-blocking
service scheduling
512MB 512MB
180G per slot, smoothly upgrade
512MB 512MB
Update of switching capacity when
user upgrade their network
CLOS architecture
512MB buffer
10GE 24GE
512M buffer for each 2 10GE ports
10GE
guarantees 200ms buffering to avoid
TCP retransmission storm 512MB buffer
10GE 24GE
10GE
28
29. S12500 Benefits
Ingress Line card
512MB
Cache
512MB Egress Line card Egress line card: Easy to cause blocking
缓存
512MB when more than one ingress line cards
Cache 512MB transmit data to one egress line cards
512MB Cache at the same time with large burst traffic
缓存
512MB 512MB Distributed buffering can handle the large
Cache Cache
512MB blocking
Burst of each interface and avoid blocking
缓存 At the egress interface
512MB
Cache
512MB
48-port GE copper/fiber full wire speed
4/8-port 10GE full wire speed (XFP) 48-port GE copper 48-port GE fiber
8/16-port 10GE full wire speed (SFP+)
32-port 10GE (SFP+)
8-port 10GE 16-port 10GE 32-port 10GE
29
30. S10500 Product introduction
Introducing the NEW S10500 core switch
• Enables 2-Tier Architectures
• 128 wire speed 10 GbE (for 8-slot
chassis)
• IRF creates a 1536 GE-port core (4
chassis stacking)
• 3 microsecond latency
• 40G, 100G ready
S10504 S10508V S10508
New generation switches, high-density 10-Gigabit and ultra high-
density Gigabit applications, future 40G, 100G ready
The H3C S10500 is engineered for:
– Core and distribution layer
– Core layer of large enterprise, data centre and campus networks
30
31. S10500 benefit-Industry leading Architecture
Fabric
The forwarding plane is physically separated
板
交换网
Fabric
from the control plane, so the MPU
背
rd d
c a ar
交换网
Fabric switchover will not affect the forwarding
e in c
traffic
Lin L
Fabric
U PU Good scalability; Smooth upgrade to next-
generation 40GE and100GE interfaces
MP M
Fabric redundancy; When one failure, others
ca card
can take over the work
rd
Lin Line
e
Easy to upgrade bandwidth, only need to
upgrade the fabric modules
31
32. Multistage CLOS Switching Network
CLOS Architecture Benefits:
Service board Switching fabric Service board
– Simplifying large-scale switching
architecture
– Reducing possible fault points and
enhance the switching architecture
reliability
– Powerful fault defense capability
– Independent switching paths will not affect
each other. Further enhances reliability
and expandability
– Supporting recursive expanding. Capacity
The H3C S10500 adopts the CLOS expansion is unlimited in theory
multi-layered switching architecture.
The service boards and the switching
fabrics are all physically connected.
The multi-stage switching architecture implements large-capacity switching capability
32
33. S10500 benefit
CLOS multi-layered switching
Fabric architecture
Dynamic non-blocking
service scheduling
480G per slot, smoothly upgrade
Update of switching capacity when
user upgrade their network
The multi-stage switching architecture
48-port GE copper/fiber
full wire speed
48-port GE copper 48-port GE fiber
8/16/32/48-port 10GE
full wire speed (SFP+)
8-port 10GE 16-port 10GE 32-port 10GE 48-port 10GE
4-port 40GE full wire speed (QSFP)
4-port 40GE
33
34. S9500E Product Introduction
Distributed ASIC+NP+Multicore OAA (Firewall, IPS, wireless, LB, network
technology analysis and etc)
Maximum 192 10GE ports IRF2
2/4/16-port 10GE line card
48-port GE Type S9505E S9508E-V S9512E
Number of slots 5 8 (vertical) 12
Switching capacity 720G 1.44T 1.44T
Packet forwarding rate 360Mpps 576Mpps 864Mpps
Engine redundancy Supported Supported Supported
Power supply
1+1 1+1 1+1
redundancy
Maxi. number of
240 384 576
Gigabit ports
Maxi. number of 10G
16 16 16
ports per slot
Maximum number of
80 128 192
10G ports
Cabinet size 11U 22U 17U
34
35. S9500E Benefits
1Ghz CPU 10ms MSTP
failover
Processor
30ms IP FRR 20k route
背板 failover Per second
EMS
EMS
BFD Detection OAM Detection
EMS
FFDR
Provide fast
Failover and 30ms BFD
FFDR convergence failover
Fast Fault Detection and
主控板 Restoration
FFDR
处
Main board 理器 Software
Offline integrated
Board diagnosis
Processor isolation management
EMS
EMS
Embedded detection detection
Innovative Multi-Engine Design POST
Maintenance of hardware of service
faults channels
Subsystem
35
36. S9500E Benefits
OSPF
Packets of each protocol are assigned to a
single queue to avoid interference between
ARP protocols
Traffic to CPU
Traffic to CPU
...
Software Control
Policy CPU
BPDU
Traffic can be limited based on pps
ICMP
Dedicated CPU traffic control
Integrate Multiple service capability
Wireless Controller NAT Load
Security: Firewall/IPS Balancing
Services: AC, NAT, Load balancing
Firewall IPS
36
37. S7500E Product Introduction
Distributed ASIC+NP+Multicore OAA (Firewall, IPS, wireless, LB, network
technology analysis and etc)
Maximum 84 10GE ports IRF2
2/4/8-port 10GE line card S7502E S7503E S7506E(V) S7510E
48-port GE
No. of Slots 4 5 8 12
backplane bandwidth ≥400G ≥1T ≥1.6T ≥2.4T
Switching capacity 192G 480G 768G 1152G/768G
Forwarding Capacity 143Mpps 274Mpps 488Mpps 773M/488Mpps
MCU Redundancy Support Support Support Support
Power supply
1+1 1+1 1+1 1+1
redundancy
Max. GE Ports 96 144 288 480
Max. 10G Ports 4 10 16 24
37
42. S5800 Market Position
IRF2 Stacking
High performance, flexible expansion
PoE+
S5800 series
(End of Row)
Big buffer (24M)
Dual Power module and hot swappable Fan
dedicated management port
S5810 series
(Top of Rack)
24-port 10GE wire speed
IRF2 stacking
Dual Power module and hot swappable Fan
10GE card S5820X series Dedicated management port
(ToR, EoR)
I
R F
Wireless control and PoE+ IPS and Firewall IRF2 Green
42
43. S5800 Design Flexibility
Firewall Wireless IPS
Upto 4-port 10GE or 16 GE per module
or wireless controller OAA service module PoE card for PoE function
USB port Hot swappable Power module
Hot swappable
Fan
43
44. S5800 Benefits-HW design
Redundant Power module
How swappable Fan Design
Dedicated management port
HW design
Green Security NetStream
BFD Multi-Core
IRF2 OAM
…. Dedicated session
Management Reliability for specific service
Routing
High performance
Multicast
Multi-Service STP processing
….
Power of SFP+ is 1W, which
is 25% of X2 (4W) 25%, 40% of
XFP (2.5W), 17% of XENPAK
(6W)
SFP+ module supports 10GE,
it also support GE SFP module
Compared with XFP
transceiver, the Cost of SFP+ is
only 50%~80% of XFP
44
45. S5800 Benefits-Security
DMZ NAT Flow
FW control
Blade
Anti-attack State HA
detection
HW design
VPN Multi-core and session
Green Security upto 256 virtual FW
IRF2
IPS
Virus Attack Protocol Blade
Management Reliability database database database
Multi-Service
Three-in-One unified Intrusion
Protection
Portal Authentication provides a
easy way for network security
Inenterprise network, S5800 can
work as aggregation equipment, can
provide L3 Portal authentication to all
aggregation users.
45
46. S5800 Benefits-Reliability
VRRPE (Enhanced VRRP)
Master A Master B
VRRP can only work in
VIP: VIP-A
MAC: MAC-A
VIP: VIP-A
MAC: MAC-B
active/standby mode, VRRPE
can work in load sharing
HW design GW IP: VIP-A GW IP: VIP-A
GW MAC: MAC-A GW MAC: MAC-B mode, more flexible, higher
performance
Green Security
Smart Link Dual uplink Protection
IRF2
Backup Link
B S9500E/7500E
Can be used to replace STP
ng
Active Link ardi
Management
F o rw c in dual-homing network
Reliability Block tr
affi
ing
S5800 Failovertime can be less
Metro Ethernet
Network than 50ms
Multi-Service A Block
ing Forwa Loading sharing among
traf r d in
f ic g
Backup Link
C S9500E/7500E dual uplinks via different
Active Link
instances
Master Transit RRPP (Rapid Ring Protection
Protocol)
Ring solution with less than
50ms failover time
Ring
Loading sharing among
different links via different
S5800
X instances
46
47. S5800 Benefits-Multi-Service
IPv6
Routing-IPv6 Static/ RIPng/OSPFV3/BGP+/IS-
ISv6
Multicast- MLD/MLD Snooping/PIM6/IPv6
multicast vlan
Firewall Tunneling-Manual/Auto 6to4/ISATA
ACL-IPv6 ACL
IPv6 management: IPv6 SNMP/TELNET/ICMPv6
Wireless
IPS/Anti-
Controller
virus
Wireless Controller Wireless Controller
Default 32 AP Default 64 AP
…..
Maximum: 64 AP Maximum: 256 AP
PoE+ (802.3AT)
S5800 series supports PoE+
Each port can provide upto
30W
Can be used to power 802.11n
AP, Video phone and etc
47
48. S5800 Benefits-Management
NTA (Network Traffic Analyzer)
S5800 can support sFlow and
Netstream
Together with iMC (NTA), S5800
HW design
NTA (Network helps customer to analyze
Green Security Traffic Analyzer) network in details
BFD (Bidirectional Forwarding
IRF2 Detection)
Can be used to detect link
Management Reliability OSPF neighbors
failover, default handover time
BFD neighbors is 10ms
Multi-Service BFD can be used with
OSPF/VRRP/BGP/ISIS/STATIC
ROUTING, failover time is less
than 50ms
SPAN/RSPAN/ERSPAN
SPAN: local switch port
l mirroring and monitoring
ne
E tun RSPAN: cross switch port
GR mirroring and monitoring
ERSPAN: cross WAN port
mirroring and monitoring via
GRE tunnel
48
49. S5800 Benefits-Green
Comply with RoHS standard
H3C complies with restriction of Hazardous
Substances (RoHS) directives that restrict the
use of certain hazardous substances in
electrical components
HW design H3C also comply with China Ministry of
Information Industry Order of Pollution,
Green Security referred to China-RoHS.
IRF2
Comply with WEEE standard
Management Reliability
H3C enforces guidelines for removing
hazardous substances and comply with
Multi-Service WEEE (Waster Electrical and Electronic
Equipments)
WEEE
S5800 Green feature
Low power mode for all unused
ports, reduce power
Modify power consumption
according to port speed
Add temperature sensor, modify fan
speed, reduce power consumption
PoE power control based on time-
slot
49
50. 1. Current Network Challenges
2. H3C IRF2 Introduction
3. H3C IRF2 Product Introduction
4. H3C Green IT
5. H3C IRF2 Case Study
50
51. H3C Green Concept
Concise design
While ensuring product functionality and performance, reduce the physical dimensions of products
to minimize the use of materials; Reduce the techniques such as electroplating, printing, and
bronzing, and use texture mapping instead; Simple package design to minimize material use;
Re-use
Greater durability of products; Modular design and component standardization to reduce frequency
of structure change;
Recycling
Use recyclable materials and avoid the use of hazardous substances; Use homogenous materials,
and mark each type of material and its lifetime to facilitate recycling;
51
52. Power Consumption Saving Design
Using the advanced chip-sets that is superior Low power mode on unused ports.
energy efficiency. Control port power depend on port speed.
Using the advanced power supply module that is Disable PoE power by time schedule.
superior energy efficiency.
Using Intelligent Cooling System
Control fan rotation speed depend
on the environment temperature
Disable power at night
And holiday
Disable power on
Unused ports
52
53. Power Consumption Comparison
FE switch
S3100-26TP-EI 17w 158% 27W C2960-24TC
GE switch
S5100-24P-EI 46w 156% 72W C2960G-24TC
10GE
158%
switch 241%
S5500-28C-EI 110w 265W C3750E-24TD
Mini 158%
Chassis S5800X-28C 245W (max) 144%353W (max) C4900M
Chassis 158%
S7500E 633W* 303% 1917W* C6500E
* With 8 10GE and 96GE hardware configuration
53
54. Low Power consumption certificate
-S5500EI
H3C S5500-EI uses 42% less Watts/Gbps when
compared to the industry average.
54
55. Low Power consumption certificate
-S7500E
H3C S7500E uses 53~55% less Watts/Gbps when
compared to the industry average.
55
56. 1. Current Network Challenges
2. H3C IRF2 Introduction
3. H3C IRF2 Product Introduction
4. H3C Green IT
5. H3C IRF2 Case Study
56
57. H3C boosts MTR
心緊生活每一
程
Caring for life’s journeys
Hong Kong’s MTR Corporation has built one of the
world's leading railway systems, well known for its
safety, reliability, customer service and cost efficiency.
H3C designed a H3C 100G future-proof information
superhighway linking 38 stations on the MTR Island,
Tsuen Wan and Kwun Tong Line, plus digital media
services along Urban Line and a new Resilient Packet
MTR Ring (RPR) upgrade along the Airport Express and
Tung Chung Line. 57
58. H3C Boosts HP Data Centre in Huston
Huston data centre is one of six major
regional HP data centers around the world.
H3C helped HP build a new flexible open
standard network infrastructure. Achieved a
“Cisco-Free” data centre environment. The
new data centre now double the port density
and capacity while reducing energy
consumption by 50%.
Products: S12500, SR88, S95E, S5820X 58
59. H3C Boots DreamWorks Make Movie
Magic
Dreamworks has deployed H3C12500 and other related
products in their data centre. With H3C’s iMC,
Dreamworks can manage its own network equipment
from other vendors into one single platform which
simplify the future network management for the data
centre. This new H3C-based network helps
Dreamworks create more innovative digital movies and
support for future business growth.
(Shrek the Fourth) 76TB of data rendered stably on
H3C-based network platform ! 59
60. H3C Supports Tencent Data Centre
Tencent currently ranked as Internet industry’s
3rd most valuable company , followed by
Google and Amazon. The company has more
than 100 million concurrent online subscribers.
H3C ‘s fast response and professional support,
reliable products and supply chain which made
H3C became Tencent’s “Grade A” network
application supplier.
60
61. Wilson Cheung
Senior Sales Manager
(852) 2501 1154
wilson_cheung@h3c.com
H3C cares your so
much, H3C wants to
know you more.
61
Editor's Notes
IRF 系统虚拟化为一台逻辑设备 逻辑设备由 1 主控板、 N 备用板和多个接口板组成 主控板管理和控制整个 IRF 系统 主控板和备用板保持配置和运行状态同步,实现 1:N 备份 当主控板工作异常时, IRF 将选择其中一台备用板成为新的主控板,保证高可靠性
So what benefits we will get from H3C solution? Compared with other vendors H3C intelligent resilient framework solution enable a flat, low latency network design, which can directly interconnect hundreds of virtual machine at the edge of network, eliminating unnecessary network hops, reduce latency and optimizing performance for high volume server to server traffic flows. H3C data center switches, together with IRF, priovide a flat, low latency, high performance, extremely reslient and easy to manage data center networks. With IRF, multiple switches are can stacked and virtualized as single unit logically (that means via only IP address management and single configuration file), H3C leverages best-in-class switch virtualization technology for core switches as well, improving scalabity and delivering unified high performance data center. As you can see from the left side, after virtualization, MSTP is eliminited, only link aggregation enabled network will bring customer mili-seconds level resilience. Before, too many configuration file will make network management and administration work even worse, now with the help of H3C IRF, one stacking unit has only one configuration file. Lastly, network is not active/standby anymore, we are creating a active/active scenario to make full use of all network resources and paths to increase the return on investment.
As we discussed before, within one data centers, H3C use IRF2 to build optimized infrastructure catering for VM scenario. However, enables distributed workloads and replicating data and applications across mutiple, geographically-dispersed data centers are a challenge. Conventional L3 solutions can not meet the high performance and low latency required VM scenario. Above all, a large L2 network is required for VM migration acorss data centers. As we understand, any single point of failure or error means loosing money for B&F, what we going to do is creat a solid platform to create more money. In order to address those issues, H3C provides superior solution that extends L2 domain with superior flexibility, high resilience and easy management. Today we will share two solutions. The first solution is 4-in-1 solution, the newly relased 100G CLOS architecture 10500 core switch is the perfect example of it. It is the first in the industry to use 4 chassis IRF virtualization, which means IRF can be extended across long distance 10GE dark fiber to bring 10500 located among multiple sites into s single giant core switches. Miliseonds level reliability is chineved natively.
Network management plays a very important role to relieve the burden of administrators and reduce TCO. H3C IMC (intelligent management center) is one of few solution in the industy which can unify mumtiple sites, multiple vendors, physical and virtual network resources via a single platform. It can help to overcome the challenges of maintaining the new virtual server edge, reduce troubleshooting for virtual networks. Therefore, IMC totally simplifies the new cutting edge data center operation and management. boosting network visibility and availability.
14 series, and nearly 200 products with the most comprehensive Ethernet switch product line in the world; Full series of core to edge switches on a virtual switching platform, leverage IRF2 technology; H3C S5500-EI and S7506E earned Miercom’s Green Certification, proving that the two save up to 40% more power than comparable models based on Industry Average. H3C S7506E 和 S55500-EI 的节能特性,经过 Miercom 绿色认证测试技术的严格评估,成功通过了 Miercom 绿色认证,无愧为高效率、符合环境保护要求的网络和数据中心解决方案。 H3C 能够提供业界覆盖最全面的交换机产品。产品包括 14 个系列,多达 200 种产品。产品囊括从园区到数据中心、盒式到箱式、从 FE 、 GE 到 10G 和 100G ,从 L2 到 L4/7 ,从 IPv4 到 IPv6 ,从接入到核心,用户都有最丰富选择和灵活组合。 在以太网领域,经历多年的耕耘和发展, H3C 积累了大量业界领先的知识产权和专利,并且全部通过中国信息产业部、 Tolly Group 、 MetroEthernet 论坛以及 IPv6 Ready 等权威部门的测试和认证,凭借丰富的产品系列和深厚的技术积累, H3C 交换机产品在数据中心、政府、金融、交通、能源、教育、企业、 SOHO 均有广泛成熟稳定的应用。 S12500 是中国国内第一款 100G 平台交换机,支持未来 40GE 和 100GE 以太网标准,整机可以提供 576 个万兆端口,提供业界密度最高的万兆接入能力;面对下一代数据中心突发流量,创新的采用了“分布式入口缓存”技术,可以实现数据 200ms 缓存,满足数据中心、高性能计算等网络突发流量的要求;为了满足数据中心级网络高可靠、高可用、虚拟化的要求, S12500 采用创新 IRF2 (第二代智能弹性架构)设计,将多台高端设备虚拟化为一台逻辑设备,同时支持独立的控制引擎、检测引擎、维护引擎,为系统提供强大的控制能力和 50ms 的高可靠保障。 S9500E 采用创新的硬件设计,通过独立的控制引擎、检测引擎、维护引擎为系统提供强大的控制能力和 50ms 的高可靠保障;通过基于分布式的高性能硬件转发和大容量 Crossbar 无阻塞交换技术,保障了系统的高处理性能和灵活的扩展能力;通过 OAA 开放架构,实现了网络和安全、无线的融合。 H3C S7500E 系列产品是杭州华三通信技术有限公司(以下简称 H3C 公司)面向融合业务网络的高端多业务路由交换机,该产品基于 H3C 自主知识产权的 Comware V5 操作系统,以 IRF2 ( Intelligent Resilient Framework 2 ,第二代智能弹性架构)技术为系统基石的虚拟化软件系统,进一步融合 MPLS VPN 、 IPv6 、网络安全、无线、无源光网络等多种网络业务,提供不间断转发、不间断升级、优雅重启、环网保护等多种高可靠技术,在提高用户生产效率的同时,保证了网络最大正常运行时间,从而降低了客户的总拥有成本( TCO )。 针对互联网数据中心关注的大流量问题, H3C 新一代互联网数据中心解决方案采用创新的分布式缓存、一次性精确调度和精细化 QoS 机制,能帮助搜狐解决突发性大流量带来的问题;同时,解决方案采用创新的 H3C 第二代智能弹性架构技术 IRF2 ,进一步简化了传统数据中心基础架构,初步实现了网络资源的虚拟化,让搜狐能够从容应对未来业务发展。 H3C 集 S12508 数据中心级核心交换机、 S75E 高端多业务路由交换机、 S5800 万兆交换机、 iMC 智能管理中心 搜狐 7*24 小时向全球提供网上贸易实时服务的需要,又具备支持近万台服务器的高密度并发高速访问的能力。因此,该项目对网络设备的可靠性和性能提出了极为苛刻的要求。包括 S12500 、 S9500E 、 S58 系列和 S55 系列在内的多款 H3C 数据中心级交换机,在淘宝网新建的大型数据中心得到了大规模的广泛应用。 美国知名媒体 Network World (网络世界)公布对 3Com 7906E/H3C S7506E 的评测结果,在经过其测试机构对该产品及 Cisco 的 Catalyst 6509 , Nexus 7010 等同类产品展开了长达 5 个月的对比评测后, H3C S7506E 在评测中得到 4.28 (满分 5 分)的高分。 Network World (网络世界)认为:“由于具有的超强性能和低能耗的特点,使得该款交换机成为大中型企业核心层和汇聚层组网的不错选择。” IRF2 功能简单的说,是把多台物理设备互相连接起来,使其成为一台统一的逻辑设备。这样无论在管理还是在使用上就成为了一个整体。也就是说,用户可以将这多台设备看成一台单一设备进行管理和使用。这样一来, IRF 可以为用户带来以下好处: 简化管理: IRF 架构形成之后,可以连接到任何一台设备的任何一个端口就以登录统一的逻辑设备,通过对单台设备的配置达到管理整个智能弹性系统以及系统内所有成员设备的效果,而不用物理连接到每台成员设备上分别对它们进行配置和管理。 简化业务: IRF 形成的逻辑设备中运行的各种控制协议也是作为单一设备统一运行的,例如路由协议会作为单一设备统一计算,而随着跨设备链路聚合技术的应用,可以替代原有的生成树协议,这样就可以省去了设备间大量协议报文的交互,简化了网络运行,缩短了网络动荡时的收敛时间。 弹性扩展:可以按照用户需求实现弹性扩展,保证用户投资。并且新增的设备加入或离开 IRF 架构时可以实现“热插拔”,不影响其他设备的正常运行。 高可靠性: IRF 的高可靠性体现在链路,设备和协议三个方面。成员设备之间物理端口支持聚合功能, IRF 系统和上、下层设备之间的物理连接也支持聚合功能,这样通过多链路备份提高了链路的可靠性; IRF 系统由多台成员设备组成, Master 设备负责系统的运行、管理和维护, Slave 设备在作为备份的同时也可以处理业务,一旦 Master 设备故障,系统会迅速自动选举新的 Master ,以保证通过系统的业务不中断,从而实现了设备级的 1:N 备份; IRF 系统有实时的协议热备份功能,负责将协议的配置信息备份到其他所有成员设备,从而实现 1:N 的协议可靠性。 高性能 对于高端交换机来说,性能和端口密度的提升会受到硬件结构的限制。而 IRF 系统的性能和端口密度是 IRF 内部所有设备性能和端口数量的总和。因此, IRF 技术能够轻易的将设备的核心交换能力、用户端口的密度扩大数倍,从而大幅度提高了设备的性能。 多业务: IRF 支持包括 IPv4 、 IPv6 、 MPLS 、安全特性、 OAA 插卡、高可用性等全部交换机特性,并且能够高效稳定地运行这些功能,大大扩展了 IRF 设备的应用范围。
December 10, 2012 HP Confidential
A Clos network is a kind of multistage circuit switching network, first formalized by Charles Clos in 1953, which represents a theoretical idealization of practical multi-stage telephone switching systems. Clos networks are required when the physical circuit switching needs exceed the capacity of the largest feasible single crossbar switch. The key advantage of Clos networks is that the number of crosspoints (which make up each crossbar switch) required can be much fewer than were the entire switching system implemented with one large crossbar switch. 100-Gbps ready platform, multi-stage, multi-plane, distributed, ClOS switching architecture, future proof design