Network Redundancy & Topologies
© 2017 Belden Inc. | belden.com | @BeldenInc
Network Redundancy – Objective
Learn the design & application details of several network redundancy protocols so you can make the best choice for your network, including a 0 packet loss redundancy method.
© 2017 Belden Inc. | belden.com | @BeldenInc
Contents
Standardized vs proprietary
Two stages of redundancy
Layer 2 redundancy
Mixing technologies
Layer 3 redundancy
The next generation
Management
© 2017 Belden Inc. | belden.com | @BeldenInc
Standardised vs Proprietary Technology Standardized: •
Works across manufacturers
•
Future-proof
•
Well understood
Proprietary: •
Fulfills niche requirements
•
In the past often with better performance and
•
More simple to configure
© 2017 Belden Inc. | belden.com | @BeldenInc
Take Note….. There is no standard for measuring network recovery time •
Exact meaning of “recovery”?
•
Network load?
•
Number of Learned Addresses?
•
Location of failure?
•
Source to destination, or round trip?
•
Type of traffic?
•
Interaction with other (redundancy) protocols?
•
…
© 2017 Belden Inc. | belden.com | @BeldenInc
Contents Standardized vs proprietary P
Two stages of redundancy
Layer 2 redundancy
Mixing technologies
Layer 3 redundancy
The next generation
Management
© 2017 Belden Inc. | belden.com | @BeldenInc
Two Stages of Redundancy Whichever redundancy method is used, there are two stages •
Re-establish the physical connection
•
Re-establish the logical/data connection
© 2017 Belden Inc. | belden.com | @BeldenInc
Re-establishing Communications PC1
A
Logical communication Physical communication Switches need to re-learn their re-established data re-established, butlogical/data NOT address tablesand for can resume logical communication communication to resume E
B
C
D Port 1
© 2017 Belden Inc. | belden.com | @BeldenInc
Switch D’s Learned Address Table Device
Path
PC1
Port 1
PC1
Port 2
PC2
Contents Standardized vs proprietary P Two stages of redundancy P
Layer 2 redundancy
Mixing technologies
Layer 3 redundancy
The next generation
Management
© 2017 Belden Inc. | belden.com | @BeldenInc
End device redundant connections •
The redundancy functionality must be provided by the end devices
•
This is not a function of the network equipment
Ethernet Network
Ethernet Network
© 2017 Belden Inc. | belden.com | @BeldenInc
Rapid Spanning Tree Protocol Objective •
Creation of resilient meshed networks
© 2017 Belden Inc. | belden.com | @BeldenInc
Graph Theory
Spanning Tree Graph
© 2017 Belden Inc. | belden.com | @BeldenInc
Rapid Spanning Tree Protocol ROOT
3
Determination of Root Bridge •
The switch with the lowest assigned priority of all the switches will be the root
•
The bridge priority default is 32768 and can only be configured in multiples of 4096 (0 being lowest value)
•
If priority value is same, hex value of MAC is used
24
92
12
4
5
7
© 2017 Belden Inc. | belden.com | @BeldenInc
Rapid Spanning Tree Protocol ROOT
Determination of Root Ports (least cost paths to root) •
Messages from any connected device to the root bridge must traverse a least cost path
•
Cost = sum of the costs of the segments on the path
•
The port connecting to that path becomes the root port (RP) of the bridge.
RP
RP
RP
RP
Assuming cost of traversing any network segment is 1 © 2017 Belden Inc. | belden.com | @BeldenInc
RP
RP
Rapid Spanning Tree Protocol ROOT
Determination of Designated Port •
•
3
Least cost path from each network segment to Root Data rate
RSTP cost
4 Mbit/s
5,000,000
10 Mbit/s
2,000,000
16 Mbit/s
1,250,000
100 Mbit/s
200,000
1 Gbit/s
20,000
2 Gbit/s
10,000
10 Gbit/s
2,000
Port connected to lowest cost path is Designated Port
DP
24
DP
92
12
DP
DP
4
DP
DP
© 2017 Belden Inc. | belden.com | @BeldenInc
5
7
Rapid Spanning Tree Protocol ROOT
3
Blocking Ports •
Any active port that is not a root port or a designated port is a blocked port
•
Port connected to lowest cost path is Designated Port
DP
DP
RP
24
RP DP
BP
12
DP
DP
RP
4
92
RP
RP DP
© 2017 Belden Inc. | belden.com | @BeldenInc
BP
5
RP
7
Rapid Spanning Tree Protocol ROOT
3
Resulting Spanning Tree Algorithm
DP
DP
RP
24
RP DP
BP
12
DP
DP
RP
4
92
RP
RP DP
© 2017 Belden Inc. | belden.com | @BeldenInc
BP
5
RP
7
Rapid Spanning Tree Protocol ROOT
3
Link Failure in Spanning Tree Network •
After link failure the spanning tree algorithm computes and spans new least-cost network tree
DP
DP
RP
24
RP DP
BP
12
DP
DP
DP RP
4
92
RP
RP RP DP
© 2017 Belden Inc. | belden.com | @BeldenInc
DP BP
5
RP
7
Things to know about the IEEE & IEC standard RSTP Recovery times
Advantages
Disadvantages
Inventor Interoperability
Recommendation
IEEE type 2sec. IEC type 5 - 20ms per hop (switch) Ring size Up to 40 switches Extremely simple to implement for basic use, but needs to be finetuned to gain reliability and faster recovery times Supports loop prevention Unpredictable recovery times (Milliseconds to Seconds) Unsuitable for large rings – maximum 40 hops Complex configuration to gain faster recovery times IEC and IEEE standardized Standardized (IEEE 802.1D-2004 & IEC62439-1) and supported across many manufacturers All switches in a RSTP network have to support RSTP If the application must tolerate multiple network failures If the customer wants a standard-based solution (but see MRP, HSR, PRP) Fast but not consistent recovery times are required Small network diameter Connection to an existing RSTP network (not exceeding RSTP specifications) © 2017 Belden Inc. | belden.com | @BeldenInc
Ring technologies Objective Creation of a reliable, highly available and resilient ring structure with predictable recovery times
© 2017 Belden Inc. | belden.com | @BeldenInc
Ring technologies “Commonalities” Recovery times
Fast recovery times with <500ms
Advantages
Predictable Recovery Times Ring size Easy to implement
>100 switches
Infrastructure
Often just one additional cable
Disadvantages
Single fault tolerant Very often proprietary implementations
Inventor
Company specific
Interoperability
Often not available
© 2017 Belden Inc. | belden.com | @BeldenInc
The Ring concept is simple.
© 2017 Belden Inc. | belden.com | @BeldenInc
Daisy-chain managed switches via any mix of copper/fiber and data speeds – up to 10Gig!
© 2017 Belden Inc. | belden.com | @BeldenInc
Assign any (one) switch to be the Redundancy Manager and close the ring RM
© 2017 Belden Inc. | belden.com | @BeldenInc
Watch-dog packets traverse entire ring, Ethernet packets only traverse active link ensuring segmentsring integrity. Watch-dog packets
RM
Ethernet data packets Link down packets
© 2017 Belden Inc. | belden.com | @BeldenInc
The Redundancy Manager activates the standby A cable or switch failure will cause Watch-dog link and instructs switches to flush/renew packets to not fully traverse the network address tables Watch-dog packets Ethernet data packets Link down packets
© 2017 Belden Inc. | belden.com | @BeldenInc
New address tables arepackets learned continue and Ethernet In parallel, Watch-dog data is automatically re-routed testing integrity Watch-dog packets Ethernet data packets Link down packets
© 2017 Belden Inc. | belden.com | @BeldenInc
Demo
© 2017 Belden Inc. | belden.com | @BeldenInc
A few incompatible ring technologies X-Ring
Ultra-Ring
N-Ring
OnTime-Ring
ICON M-Ring
Ring
P-Ring
V-Ring
HSR* Siemens
Rapid-Ring T-Ring
Rapid Super Ring
HIPER-Ring Z-Ring Real-Time Ring
FRNT S-Ring © 2017 Belden Inc. | belden.com | @BeldenInc
Comparison of Redundancy Protocols defined in IEC62439 Protocol
Most current Standard
STP
Spanning Tree Protocol
RSTP
Rapid Spanning Tree Protocol IEEE 802.1D-2004 Cross-Network Redundancy Protocol IEC 62439-4:2010
CRP BRP
IEEE 802.1d
IEC 62439-5:2010
DRP
Beacon Redundancy Protocol Distributed Redundancy Protocol
MRP
Media Redundancy Protocol
IEC 62439-2:2010
Fast MRP
Media Redundancy Protocol
30s 2s 1s worst case for 512 end nodes 4...8ms worst case for 500 end nodes 100ms worst case for 50 switches 200ms worst case for 50 switches 30ms worst case for 50 switches 10ms worst case for 15 switches
Remark any topology/mesh, diameter limited any topology/mesh, diameter limited any topology/ duplicated networks Two top level switches with star, line or ring topologies
1990 2004 2007 2007
ring, double ring
2010
ring
1998/2007
ring
2010
ring
2010
HSR
IEC 62439-2:2010 IEEE 802.1D-2004 (configuration requirements described in IEC Rapid Spanning Tree Protocol 62439-1:2010) 5...20ms per switch High-Availability Seamless Redundancy IEC 62439-3:2012-07 0ms
ring
2010
PRP
Parallel Redundancy Protocol IEC 62439-3:2012-07
any toplogy/ duplicated networks 2010
Optimized RSTP
(1) (2)
IEC 62439-6:2010
Typical re-config
Available since
0ms
pre-standard Hiper Ring since 1998, MRP since 2007 pre-standard Fast Hiper Ring since 2007
© 2017 Belden Inc. | belden.com | @BeldenInc
IEC62439 Redundancy Standardized meshed, ring and back-up line topologies
MRP
Fast MRP
200ms optimized 80ms
10ms
5-20ms per hop RSTP
© 2017 Belden Inc. | belden.com | @BeldenInc
Device level ring (DLR) Standardized ring topologies
DLR
Sub ms
Defined in the ODVA Combinations with other redundancy technologies possible
© 2017 Belden Inc. | belden.com | @BeldenInc
Things to know about the “Industrial Standard” HIPER Ring Recovery times
Standard Mode 200ms Fast HIPER Mode 40ms
Advantages
Predictable Recovery Times Ring size Up to 200 switches >20.000 switches Fast Hiper-Ring Extremely simple to implement
Disadvantages
Proprietary and single fault tolerant
Inventor
Hirschmann launched a Hub-based ring in 1990. This was the first version of “HIPER-Ring”
Interoperability
Only if all devices support HIPER Ring Not possible to reset Learned Address Tables Unpredictable recovery time Multicast watchdog packets will be broadcast
Recommendation
Upgrade of existing (HIPER Ring) applications © 2017 Belden Inc. | belden.com | @BeldenInc
Things to know about the IEC standard MRP Recovery times Advantages
Disadvantages Inventor
Interoperability
Recommendation
Standard Mode 80ms Fast MRP Mode 30ms Predictable Recovery Times Ring size Up to 200 switches Extremely simple to implement Single fault tolerant MRP is an IEC standardized version of HIPER Ring, with some optimizations As an IEC standard, any manufacturer can implement MRP. Hirschmann has added some additional features: Support for ring coupling Maximum 200 switches in a ring (IEC standard = 50) 80ms recovery time (standard 200ms) Application requires consistent recovery times Customer wants a clear network topology Geography is not suitable for a meshed structure Minimized commissioning and maintenance effort Customer wants a standard based solution © 2017 Belden Inc. | belden.com | @BeldenInc
Setting up MRP using Automatic Ring Configuration & Diagnostic (ARC) • Connect all “out of the box” switches to form a ring topology • Select any (one) switch to be the Redundancy Master and assign it an IP address • User web browser or HiView to enter that switch’s management • Run ARC • ARC checks switches, verifies ring topology, automatically enables MRP and saves configurations
© 2017 Belden Inc. | belden.com | @BeldenInc
Things to know about the ODVA standard DLR Recovery times Advantages
Disadvantages Inventor Interoperability
Recommendation
50 nodes <3ms Predictable Recovery Times Ring size Up to 250 switches Easy to implement Single fault tolerant Published in ODVA EtherNet/IP specification, November, 2008 As an ODVA standard, any ODVA member can implement DLR.
Application requires consistent and very fast recovery times Customer wants a clear network topology Geography is not suitable for a meshed structure Minimized commissioning and maintenance effort Customer wants a standard based solution
© 2017 Belden Inc. | belden.com | @BeldenInc
Multiple Rings
Possibilities RM
SRM1
Base-Ring SRM1
Sub-Ring 1 SRM2
SRM2
Sub-Ring 2
RM – Redundancy Manager SRM – Sub-Ring Manager
© 2017 Belden Inc. | belden.com | @BeldenInc
Multiple Rings
Possibilities SubRing SubRing
RM
SRM
Sub-Ring
Base-Ring SRM
SubRing
SubRing
SubRing
© 2017 Belden Inc. | belden.com | @BeldenInc
RM – Redundancy Manager SRM – Sub-Ring Manager
Multiple Rings
Possibilities
RM
SRM SRM
Base-Ring
SRM SRM
RM – Redundancy Manager SRM – Sub-Ring Manager
© 2017 Belden Inc. | belden.com | @BeldenInc
Multiple Rings
Possibilities RM
SRM
Base-Ring SRM
SRM
Sub Ring SRM
Sub Ring
RM – Redundancy Manager SRM – Sub-Ring Manager
© 2017 Belden Inc. | belden.com | @BeldenInc
Multiple Rings
RM
Possibilities
Base-Ring SRM
SRM RM – Redundancy Manager SRM – Sub-Ring Manager
© 2017 Belden Inc. | belden.com | @BeldenInc
Things to know about Multiple Rings Recovery times Advantages
Disadvantages Inventor Interoperability
<100ms Predictable Recovery Times Ring size Up to 200 switches (MRP) Easy to implement tolerates multiple network failures (depending on location) up to 16 sub-rings are possible Proprietary technology Multiple vendors have developed proprietary technologies No
Requirements
Sub Ring starts and ends at the same Base (backbone) Ring Base Ring has to be either HIPER Ring or MRP
Recommendation
Application requires consistent and very fast recovery times Customer wants a clear network topology Minimized commissioning and maintenance effort
© 2017 Belden Inc. | belden.com | @BeldenInc
Ring Extension: Ring/Net Coupling
Redundant Connection of Multiple Rings or Networks
RM
• This allows the redundant coupling of redundant rings and network segments. • Two rings or network segments (or multiple combinations of the two) are connected via two separate paths.
Base-Ring
Master Coupling Port Active
Slave Coupling Port on Standby
Base-Ring
RM
© 2017 Belden Inc. | belden.com | @BeldenInc
Redundant Ring/Net Coupling
Compensates two faults in a Ring topology
© 2017 Belden Inc. | belden.com | @BeldenInc
Things to know about redundant Ring-/ Net-coupling Recovery times Advantages
<250ms
Disadvantages
Predictable Recovery Times Simple to implement Can compensate for two faults in a Ring topology Proprietary technology
Interoperability
No
Recommendation
Application requires consistent and fast recovery times Redundant coupling of redundant rings and/or network segments. Customer wants a clear network topology Minimized commissioning and maintenance effort
© 2017 Belden Inc. | belden.com | @BeldenInc
Link Aggregation Objective •
Create a single high-bandwidth logical link from multiple lowerbandwidth physical links (trunking)
•
Not developed for redundancy, but to increase bandwidth
1x 1Gigabit = 4Gig 2x 3x 4x 1Gig 2Gig 3Gig
Link Aggregation Control Protocol (LACP)
© 2017 Belden Inc. | belden.com | @BeldenInc
Link Aggregation over wireless
Wireless LAN 2.4 GHz WLAN
Wireless LAN 5 GHz WLAN
© 2017 Belden Inc. | belden.com | @BeldenInc
Link Aggregation Advantages •
Standardized 802.1ax and 802.1aq
•
Easily configured
•
Fast recovery times
•
Increased throughput
•
Commonly used in IT to increase bandwidth to servers
Disadvantages •
Unpredictable recovery times
© 2017 Belden Inc. | belden.com | @BeldenInc
Link Backup – Link Redundancy Link Interface pair consist of any combination of physical interfaces, e.g. one 100Mbit SFP and the other 1000Mbit One is the standby link, the other the primary (active). Only the active port is forwarding traffic. If the primary link shuts down, the standby link starts forwarding traffic.
Example: ports 1 and 2 on switch A are connected to uplink switches B and C. If port 1 is the active link, communication is enabled. Port two is in stand by mode. If link 1 goes down, port 2 will be enabled and starts forwarding traffic to switch C. If link 1 comes back again, communication will still go through port 2. Port 1 is now the new stand by port.
B
C
Port 2
Port 1
© 2017 Belden Inc. | belden.com | @BeldenInc
A
Link Backup Advantages •
Easily configured
•
Fast recovery times B
C
Disadvantages •
Proprietary
Port 2
Port 1 A
© 2017 Belden Inc. | belden.com | @BeldenInc
Contents
PStandardized vs proprietary PTwo stages of redundancy PLayer 2 redundancy
Mixing technologies
Layer 3 redundancy
The next generation
Management
© 2017 Belden Inc. | belden.com | @BeldenInc
Mixing technologies When possible, do not mix standardized and proprietary Layer 2 redundancy technologies •
Integrity/network health packets (watch-dog and similar) may not be allowed to pass
•
Switches may not know that they need to flush and re-learn their address tables
Proprietary
Standard Standard and Proprietary
© 2017 Belden Inc. | belden.com | @BeldenInc
MRP (Media Redundancy Protocol) over Link Aggregation Combination of Link Aggregation and MRP To increase the availability or bandwidth for some connections or the entire ring MRM
Link Aggregation
Example: MRP over LAG for one connection only MRM
Example: MRP over LAG for the entire ring
© 2017 Belden Inc. | belden.com | @BeldenInc
Combining MRP and RSTP Using RSTP on MRP In the MRP compatibility mode, the device allows to combine RSTP with MRP.
MRP
With the combination of RSTP and MRP, the fast switching times of MRP are maintained. RSTP
RSTP applies to the devices outside the MRP-Ring.
© 2017 Belden Inc. | belden.com | @BeldenInc
Redundancy combinations with DLR and Link aggregation
DLR
© 2017 Belden Inc. | belden.com | @BeldenInc
Redundancy combinations with DLR and Link backup
DLR
© 2017 Belden Inc. | belden.com | @BeldenInc
Redundancy combinations with DLR and MRP
DLR
MRP
© 2017 Belden Inc. | belden.com | @BeldenInc
Redundancy combinations with DLR and MRP/Link aggregation
DLR
MRP
© 2017 Belden Inc. | belden.com | @BeldenInc
Redundancy combinations with DLR and RSTP
DLR
RSTP
© 2017 Belden Inc. | belden.com | @BeldenInc
Redundancy combinations with DLR and RSTP
DLR
RSTP
© 2017 Belden Inc. | belden.com | @BeldenInc
Redundancy combinations with DLR and Sub-Ring
DLR
Sub-Ring
© 2017 Belden Inc. | belden.com | @BeldenInc
Contents Standardized vs proprietary P Two stages of redundancy P PLayer 2 redundancy Mixing technologies P
Layer 3 redundancy
The next generation
Management
© 2017 Belden Inc. | belden.com | @BeldenInc
Layer 3 Redundancy Redundancy for routers •
VRRP
•
HiVRRP
Redundancy for links •
RIP
•
OSPF
© 2017 Belden Inc. | belden.com | @BeldenInc
Virtual Router Redundancy Protocol VRRP (RFC 3768) 10.0.0.0/24 10.0.0.1
Master Hello packets each second Recovery time 3 sec.
Virtual IP address and MAC address
192.168.0.1
192.168.0.0/24 © 2017 Belden Inc. | belden.com | @BeldenInc
Backup
HiVRRP •
Proprietary
•
Same principle as VRRP
•
Faster
•
•
VRRP hello packet each second after 3 second alternative backup
•
HiVRRP hello packet each 100ms after 300ms alternative backup
10 times faster than the standard
© 2017 Belden Inc. | belden.com | @BeldenInc
Routing Protocols (link redundancy) Task: dynamic selection of paths through a network •
Static routing •
•
Manually enter the routes
Routing protocols •
RIP
Routing Information Protocol
•
OSPF
Open Shortest Path First
•
(BGP
Border Gateway Protocol)
WAN
•
(IGRP
Interior Gateway Routing Protocol)
WAN
•
….
© 2017 Belden Inc. | belden.com | @BeldenInc
RIP (Distance Vector) Router Information Protocol (RFC 1058, v2: RFC 1723) •
For networks with a small number of routers •
Only metric is hop count
•
Maximum 15 hops
© 2017 Belden Inc. | belden.com | @BeldenInc
RIP Routing Tables
Stuttgart A 1 hop B 1 hop C 2 hops D 2 hops E 2 hops F 2 hops G 3 hops H 3 hops
N/A N/A Frankfurt Frankfurt Munich Frankfurt Frankfurt Frankfurt
Exchanging route information
© 2017 Belden Inc. | belden.com | @BeldenInc
OSPF (Link State) Open Shortest Path First (RFC 1247) •
For networks with a large number of routers
•
Link state algorithm with criteria (metrics) for routing decisions: •
Path cost
•
Type Of Service field (prioritisation)
© 2017 Belden Inc. | belden.com | @BeldenInc
Shortest Path First Algorithm
Munich 0
Stuttgart 188
Frankfurt 305
Hannover 585
© 2017 Belden Inc. | belden.com | @BeldenInc
Leipzig 367
Berlin 512
Comparison - Dynamic Routing Protocols Distance Vector Routing (RIP) •
Small networks
•
Simple
•
Slow in noticing changes (convergence), approx. 180sec
Link State Routing (OSPF) •
Large networks
•
Complex
•
Fast in noticing changes, approx. 30 sec.
© 2017 Belden Inc. | belden.com | @BeldenInc
Contents
PStandardized vs proprietary PTwo stages of redundancy PLayer 2 redundancy PMixing technologies PLayer 3 redundancy
The next generation
Management
© 2017 Belden Inc. | belden.com | @BeldenInc
A few incompatible ring technologies X-Ring
Ultra-Ring
N-Ring
OnTime-Ring
ICON M-Ring
Ring
P-Ring
V-Ring
HSR* Siemens
Rapid-Ring T-Ring
Rapid Super Ring
HIPER-Ring Z-Ring Real-Time Ring
FRNT S-Ring © 2017 Belden Inc. | belden.com | @BeldenInc
Comparison of Redundancy Protocols defined in IEC62439 Protocol
Most current Standard
STP
Spanning Tree Protocol
RSTP
Rapid Spanning Tree Protocol IEEE 802.1D-2004 Cross-Network Redundancy Protocol IEC 62439-4:2010
CRP BRP
IEEE 802.1d
IEC 62439-5:2010
DRP
Beacon Redundancy Protocol Distributed Redundancy Protocol
MRP
Media Redundancy Protocol
IEC 62439-2:2010
Fast MRP
Media Redundancy Protocol
30s 2s 1s worst case for 512 end nodes 4...8ms worst case for 500 end nodes 100ms worst case for 50 switches 200ms worst case for 50 switches 30ms worst case for 50 switches 10ms worst case for 15 switches
Remark any topology/mesh, diameter limited any topology/mesh, diameter limited any topology/ duplicated networks Two top level switches with star, line or ring topologies
1990 2004 2007 2007
ring, double ring
2010
ring
1998/2007
ring
2010
ring
2010
HSR
IEC 62439-2:2010 IEEE 802.1D-2004 (configuration requirements described in IEC Rapid Spanning Tree Protocol 62439-1:2010) 5...20ms per switch High-Availability Seamless Redundancy IEC 62439-3:2012-07 0ms
ring
2010
PRP
Parallel Redundancy Protocol IEC 62439-3:2012-07
any toplogy/ duplicated networks 2010
Optimized RSTP
(1) (2)
IEC 62439-6:2010
Typical re-config
Available since
0ms
pre-standard Hiper Ring since 1998, MRP since 2007 pre-standard Fast Hiper Ring since 2007
© 2017 Belden Inc. | belden.com | @BeldenInc
IEC62439 Redundancy When fast is not fast enough
MRP
Fast MRP
200ms optimized 80ms
10ms
5-20ms per hop RSTP
© 2017 Belden Inc. | belden.com | @BeldenInc
IEC62439 Redundancy PRP – (Parallel Redundancy Protocol)
Zero failover – duplicated networks
PRP Sender Red Box 1
Lan A
No packet loss
PRP Receiver Red Box 2
Lan B Port A Port B
Two redundant networks By doubling the packets no data loss if one packet fails PRP-Redundancy-Box = bidirectional splitter and combiner
© 2017 Belden Inc. | belden.com | @BeldenInc
IEC62439 Redundancy PRP – (Parallel Redundancy Protocol)
Zero failover – duplicated networks
No packet loss
SAN
SAN
SAN
SAN
SAN
SAN
© 2017 Belden Inc. | belden.com | @BeldenInc
IEC62439 Redundancy PRP – (Parallel Redundancy Protocol)
Zero failover – duplicated networks
No packet loss
SAN
SAN
SAN
SAN
SAN
SAN
© 2017 Belden Inc. | belden.com | @BeldenInc
IEC62439 Redundancy PRP – (Parallel Redundancy Protocol)
Zero failover – duplicated networks
No packet loss
example standard LAN with RSTP
SAN
SAN
SAN
SAN
SAN
SAN
© 2017 Belden Inc. | belden.com | @BeldenInc
IEC62439 Redundancy PRP – (Parallel Redundancy Protocol)
Zero failover – duplicated networks
No packet loss
example standard LAN with MRP
SAN
SAN
SAN
SAN
SAN
SAN
© 2017 Belden Inc. | belden.com | @BeldenInc
IEC62439 Redundancy PRP – (Parallel Redundancy Protocol)
Zero failover – duplicated networks example standard LAN with MRP
MRP
MRP
© 2017 Belden Inc. | belden.com | @BeldenInc
No packet loss
IEC62439 Redundancy PRP – (Parallel Redundancy Protocol) No packet loss
Zero failover – duplicated networks example standard LAN with wireless
BAT-R
BAT-R
Wireless LAN
SAN
SAN
SAN
SAN
SAN
MRP
© 2017 Belden Inc. | belden.com | @BeldenInc
SAN
IEC62439 Redundancy PRP – (Parallel Redundancy Protocol) No packet loss
Zero failover – duplicated networks example standard LAN with wireless
Wireless LAN BAT-R
2.4 GHz WLAN BAT-R
SAN
SAN
SAN
SAN
SAN
Wireless LAN 5 GHz WLAN
© 2017 Belden Inc. | belden.com | @BeldenInc
SAN
IEC62439 Redundancy PRP – (Parallel Redundancy Protocol) No packet loss
Advantages of PRP over WLAN a) Compensation of lost packets instead of compensation in case of NW failure (per-packet basis) b) Reduced latency and jitter, because PRP always forwards the faster packet
© 2017 Belden Inc. | belden.com | @BeldenInc
Effects of PRP over WLAN – Packet loss WLAN A WLAN B
PRP •
Packet loss without PRP
Correlated losses
Directly visible to application •
Packet loss with PRP • •
•
•
Use of duplicate packets Application only experiences loss if both packets are lost at the same time Loss rate for uncorrelated losses Verluste: V1xV2 = VPRP E.g., 1% x 1% = 0,01% (100 times better)
Challenge: Losses on both channels must be unrelated •
•
Use of different channels/frequency bands diversity Elimination of common sources of losses © 2017 Belden Inc. | belden.com | @BeldenInc
Effects of PRP over WLAN – Latency and Jitter WLAN A
WLAN B
PRP
•
•
Root cause of high jitter and latency in WLANs •
Busy channel (CSMA)
•
Layer 2 retransmissions because of interference / bad SNR
Positive effects of PRP in WLANs •
Jitter and latency only increase if the packets on both paths are delayed
•
PRP always delivers the faster packet
Always: as low or better latency and jitter than the better path
© 2017 Belden Inc. | belden.com | @BeldenInc
HSR (High Available Seamless Ring)
PRP Packet
PRP Packet
© 2017 Belden Inc. | belden.com | @BeldenInc
IEC62439 Seamless Redundancy PRP & HSR – Complicated to configure? No packet loss
2 switches • 1x “click” each • Connect the ports
© 2017 Belden Inc. | belden.com | @BeldenInc
Combination of HSR and PRP Compensates multiple failures No packet loss
LAN A
LAN B
© 2017 Belden Inc. | belden.com | @BeldenInc
Contents Standardized vs proprietary P Two stages of redundancy P PLayer 2 redundancy Mixing technologies P PLayer 3 redundancy PThe next generation
Management
© 2017 Belden Inc. | belden.com | @BeldenInc
Management •
Industrial HiVision
•
Other SNMP management platforms
•
SCADA via OPC tags and ActiveX control
•
Relay contact
© 2017 Belden Inc. | belden.com | @BeldenInc
Managing Redundancy
• •
• • •
Did you lose a link? With fast reconvergence, how do you know? Industrial HiVision or other SNMP management platforms SCADA via OPC tags and ActiveX control Using Industrial Profiles, Direct integration into RSLogix 5000 or SIMATIC STEP 7 Relay contact to IO or other device
© 2017 Belden Inc. | belden.com | @BeldenInc
Summary Implement redundancy only if required Use if possible standardized technologies First choice RSTP no configuration • •
Smaller, non-critical networks Fast but not consistent recovery times
MRP or DLR •
Larger and/or mission-critical networks • Predictable and fast recovery times
PRP/HSR •
Small and/or larger networks • Best available solution for mission critical and uninterrupted communication © 2017 Belden Inc. | belden.com | @BeldenInc
Demo
© 2017 Belden Inc. | belden.com | @BeldenInc
© 2017 Belden Inc. | belden.com | @BeldenInc