Upload
truongkiet
View
217
Download
0
Embed Size (px)
Citation preview
Design Rules for Mission Critical Networks
OTN Systems
Who is OTN Systems?y• OTN Systems develops and supplies fiber optic
i i k f i i lcommunication networks for critical infrastructures such as Public Transit Systems, R il Oil & G i li d MiRailways, Oil & Gas pipelines and Mines.
• More than 15 years of experience with installations world-wide: –US: Amtrak, SEPTA, San Diego Sprinter, Las Vegas
Monorail, …
3 design rules for mission critical networksnetworks
Keep it simpleKeep it simpleKeep it separated
• Following these basic rules in network design
Make it reliableFollowing these basic rules in network design will result in networks that are easier to configure troubleshoot and maintainconfigure, troubleshoot and maintain.
• The end result is a reliable network with low operational costsoperational costs.
Simplicityp y
Keep it simple: Why?p p y• Operational simplicity is a key element to keep
i l loperational cost low– Simplification of network design & planning– Speed-up troubleshooting and maintenance– Project risk reduction by lowering project
complexity– Reduction of training costs / knowledge
b lobsolescence– Avoidance of human error
Keep it simple: How?p p• Work at the lowest network layer possible
(L1/L2/L3)(L1/L2/L3).• Use hot-swap modular hardware.• Use system-wide graphical network
management tools, preferably with g p yconfiguration wizards.
• Separate the network into smaller, easy toSeparate the network into smaller, easy to manage application sub-networks (see rule #2).
Simplicity: L1/L2 where possible, L3 where requiredwhere required
• As networks use higher layers of the OSI model, they become , ymore complex.
• Provide the required features at the lowest complexity possiblethe lowest complexity possible– Hardware protection and Network
protection at L1 if possible• no RSTP (L2) or OSPF (L3) needed
– Each network segment should be able to work at L2 or L3 as required qby the individual application
– Non-Ethernet & IP applications (e g serial data analog voice) can(e.g. serial data, analog voice) can be connected at L1 (native interface) without port servers.
The dilemma: Layer 2 or Layer 3?y yL2 EthernetNG-Sonet
• L2: Switches– Filtering based on MACg– No broadcast
suppression– Limited complexity
B t ll t it bl f• But usually not suitable for large multiservice networks
• L3: RoutersL3 routed network • L3: Routers– Filtering based on IP– Broadcast suppression
P l d d– Protocols needed to control traffic flows in the network
– Complexity is highComplexity is high
Simplicity: L2 where possible, L3 where requiredwhere required
Instead of one large complex L3 network, the network is divided into smaller independent application networks that can each run at L2 or
6. Presentation
7. Application L3 as required.•Smaller independent networks are easier to control
Decide for each individual application SLAN if it operates at L2 or L3•Overall complexity is reduced by mix of L2&L3
4. Transport
5. Session
Overall complexity is reduced by mix of L2&L3•Choose to tunnel or peer L2/L3 control protocols
ng syst
ems
ddre
ss
AN e sig
ns
ncy
voic
e
cont
rol
2 Data link
3. Network
Sign
allin
SCAD
A
Cont
rol
Publ
ic A
Offi
ce L
A
Mes
sage
CCTV
Emer
gen
Acce
ss c
1. Physical
2. Data link
Keep things separatedp g p
Separation: why?p y• In multi-service networks different applications are
combined on a single network infrastructurecombined on a single network infrastructure.• For each application it should be as if it runs on a
separate dedicated networkseparate dedicated network.– In this way each application network can be treated
independently.
• No mutual interference between applications may occur.– E.g. bandwidth contention when new applications are added– E.g. a faulty application that disturbs the other network based
li tiapplications
• Rogue users should remain isolated.
Separation: how?p• A dedicated amount of bandwidth is allocated
h li i i h kto each application or service on the network, which provides hard QoS for each service
• Ports that must be able to communicate are configured in a network service with a reserved amount of bandwidth (point-to-point or multipoint SLAN).
• L3 routing between the application networks is possible if required.p q
Separation by providing dedicated & guaranteed bandwidth to each application (Hard QoS)pp ( Q )
Network capacity
T l h V IPTelephony-VoIP
WiFi access points
Each application gets a dedicated amount of bandwidth.Bandwidth is always
SCADA (e.g. Profinet)
Administration LAN
Bandwidth is always available: 100% QoSNo unwanted interference between applications
Operations/Control LAN
10Gbps applications.Bandwidth allocation is managed by the Management System.Bandwidth is used
IP Video Surveillance
Bandwidth is used efficiently thanks to flexible allocation.Network management communication uses acommunication uses a dedicated channel.Network Management
Communication
Separation: Different application layers can be configured onto a single networkconfigured onto a single network.
Network Management
VOICE Network Layer
SCADA Network Layer
Video Network Layer
Layer
PhysicalN50/N70
Physical OTN Network
Separation: Example for different Metro & Rail applicationspp
Signaling/CBTC
SCADAO ti &
Application Mapping in dedicated services
SCADA
Traffic monitoring
Control systems
Vid S ill (CCTV)
Operations & Control
Video Surveillance (CCTV)
Emergency call box
Access control
P bli Add (PA)
Security
Public Address (PA)
Passenger Information (PIS)
Public Wi-Fi AP
Information10G
Operational Telephony
Hotline
Train radioCommunication
Office LAN
Operational radio
Ticket vending
Administration
Commercial
Make it reliable: always on!y
Make it reliable: Why?y• Network has to work reliably in harsh conditions
– dust, EMC, temperature
• Network downtime has to be avoided– No operational losses, maintenance & service costs
• Network needs to have a long operational g plifetime (>15y) – No cost for replacementNo cost for replacement
• Network needs to be future proofNo cost for upgrading or replacement– No cost for upgrading or replacement
Make it reliable: How?
Availability =MTBF
Availability =(MTBF+MTTR)
• High MTBF (Mean Time Between Failures)– Network should be reliable under harsh conditions
=> industrial grade equipment– Provide hardware redundancy & network redundancy
L MTTR (M Ti T R i )• Low MTTR (Mean Time To Repair)– Network should be easy & fast to repair
=> simplicity network management tools & modular design=> simplicity, network management tools & modular design
Conclusions
Total Cost of Ownershipp
Operations & maintenance costs
(source: UITP Core Brief on Life Cycle Cost Optimization, September 2009)
Operations & maintenance costscan be greatly reduced by simplicity and reliability!
Typical Initial equipment cost
TCO Analysis Transportation Area
Conclusion: simplicity, separation and reliability guarantee low OPEXreliability guarantee low OPEX
Low opex High opex
Equipment Cost
CAPEX
OPEXMaintenance
Training
Updating
installation
Complexity
UnavailabilityUnavailability
Conclusion: what does it mean for your Company?your Company?Reliability Redundancy (in hardware L1 L2 & L3) Redundancy (in hardware, L1, L2 & L3) Long-term availability Harsh environments
Reliable service & Reduction of TCO
Separation Single network for all applications 100% QoS guaranteed for each
Simplicity Reduction of network complexity100% QoS guaranteed for each
application L2 or L3 as required per SLAN
p y User friendly management & wizards L2 where possible, L3 where required
Conclusion: what does it mean for You?
Reliability Redundancy (in hardware L1 L2 & L3)
You?
Redundancy (in hardware, L1, L2 & L3) Long-term availability Harsh environments
Peace of mind!YOU Peace of mind!YOU
Separation Single network for all applications 100% QoS guaranteed for each
Simplicity Reduction of network complexity100% QoS guaranteed for each
application L2 or L3 as required per SLAN
p y User friendly management & wizards L2 where possible, L3 where required
Contact us
OTN Systems nvIndustrielaan 17bB-2250 OlenB 2250 OlenBelgium
E-mail: [email protected]: www.otnsystems.com