CONMUTACIÓN Y RUTEO II

Clase 2. Fast Convergence &

Network Quality Analysis

Alberto Arellano A. Ing. Msc.

[email protected]
CCNA – CCNP – CCSP - LPI
Evolution Failure Handling
Network Convergence Overview
 Network convergence is the time needed for traffic to be rerouted
to the alternative or more optimal path after the network event.

 Network convergence requires all affected routers to process the

event and update the appropriate data structures used for
forwarding

 Network convergence is the time required to:

• Detect the event T1
• Propagate the event T2
• Process the event T3
• Update the routing table/FIB T4
Need for Fast Convergence
 Its not only browsing, mail and watching videos any more.
 Internet and Networks carrying Voice/Video calls,
telepresence systems.
 Carrying business and misión critical data.
 No option for outage or interruption.
 At present it takes few seconds without any fast convergence
techniques applied in Interface and protocol configuration.
Network Convergence Overview
Routing Protocol Convergence Comparison using Simulation and Real
Equipment. D. Sankar and D. Lancaster.
Centre for Security, Communications and Network Research Plymouth University,
United Kingdom. 2013
Failure Detection
 Detecting the failure is one of the most critical and often
one of the most challenging part of network
convergence.

 Failure Detection can occur on different levels/layers

• Physical Layer (1) - G.709 (IPoDWDM), SONET/SDH
(POS), Ethernet autonegotiation Transport
• Layer (2) - PPP or HDLC keepalives, Frame-Relay
LMI, ATM-OAM, Ethernet-OAM
• Network Layer (3) – IGP Hello, BFD
• Application Layer
OSPF Aggressive Timers (Fast Hellos)
 OSPF Default timers

 OSPF supports aggressive timers to decrease link

failure detection
• Timers can be tuned to a minimum of 1 second dead interval
• Number of Hello packets per seconds specified as multiplier – 3 to 20
• Interface dampening is recommended with sub-second hello timers
• OSPF point-to-point network type to avoid designated router (DR)
negotiation
Fast Hellos IGPs
Network Convergence Time
 Bidirectional Forwarding Detection - BFD
 RFC 5880 – Juniper Networks – June 2010
 Lightweight hello protocol designed to run over multiple transport
protocols
 Designed for sub-second Layer 3 failure detection
 Any interested client
• EIGRP
• IS-IS
• OSPF
• Static Routes
 All registered clients benefit from uniform failure detection
 Runs on physical, virtual and bundle interfaces
 Uses UDP port 3784 / 3785 (for echo)
 BFD can provide reaction time close to 50 milliseconds
BFD Router Architecture
BFD Protocol
Two types of hello packet:

BFD control packet : each peer needs to process the

packet.

Echo packet : when one peer receive echo packet, it

just loop back the echo to the same link without
processing packet
BFD Protocol
• BFD provides fast peer failure detection times
independently of all media types, encapsulations,
topologies, and routing protocols OSPF, BGP,
HSRP, MPLS LSP, etc.

• Some vendor product can support static route

binding and layer 2 link detection.
BFD Protocol
When BFD doesn’t receive its control packets
anymore it realizes we have a link failure and it will
report this to OSPF. OSPF will then tear down the
neighbor adjacency.
 BFD Configuration - CISCO
BFD timers are configured under each interface with the
command:
bfd interval [send-timer] min_rx [receive-timer] multiplier [number]

• The send-timer specifies the frequency of BFD packets

originated by the router
• The receive-timer the minimum interval between packets
accepted from BFD peers.
• The multiplier number is the number of BFD packets that can
be lost before the BFD peer is declared down
BFD Configuration - CISCO

7604(config)#interface TenGigabitEthernet 4/0/1

7604(config-if)#bfd interval 50 min_rx 50 multiplier 3
7604(config-if)#ip ospf bfd

7604(config)#router ospf 100

7604(config-router)#network 1.1.1.1 0.0.0.0 area 0
7604(config-router)#exit

RP/0/RSP0/CPU0:ASR9K(config)#router ospf 100

RP/0/RSP0/CPU0:ASR9K(config-ospf)#bfd minimum-interval 50
RP/0/RSP0/CPU0:ASR9K(config-ospf)#bfd multiplier 3
RP/0/RSP0/CPU0:ASR9K(config-ospf)#area 0
RP/0/RSP0/CPU0:ASR9K(config-ospf-ar)#interface TenGigE 0/0/0/0
RP/0/RSP0/CPU0:ASR9K(config-ospf-ar-if)#bfd fast-detect
RP/0/RSP0/CPU0:ASR9K(config-ospf-ar-if)#commit
 BFD Configuration – Juniper -HPE

set protocols ospf area 0.0.0.0 interface em0 bfd-liveness-detection minimum-interval 100
set protocols ospf area 0.0.0.0 interface em0 bfd-liveness-detection multiplier 3

[RouterA] interface GigabitEthernet 2/1/1

[RouterA-GigabitEthernet2/1/1] ospf bfd enable
[RouterA-GigabitEthernet2/1/1] bfd min-transmit-interval 500
[RouterA-GigabitEthernet2/1/1] bfd min-receive-interval 500
[RouterA-GigabitEthernet2/1/1] bfd detect-multiplier 7
[RouterA-GigabitEthernet2/1/1] return
Lab. OSPF with BFD
In this scenario, R1, R2 and R3 are all part of OSPF. We will simulate a
failure caused by CLOUD PROVIDER in which the interface on R1 will not
go down, hence OSPF will need to wait for the dead timer interval to
expire, which is 40 seconds by default, so it can switch over the alternate
path through R3.
OSPF without BFD
OSPF without BFD – Fail Cloud Provider
OSPF without BFD – Fail Cloud Provider

25 segundos
OSPF with BFD – Config
OSPF with BFD – Config
OSPF with BFD – Fail Cloud Provider
OSPF with BFD – Homework
R1  HPE
OSPF LFA - FRR
OSPF Loop-Free Alternate (LFA) Fast Reroute (FRR) allows OSPF to
quickly switch (within 50 ms) to a backup path when a primary path
fails. Without LFA FRR, OSPF has to re-run SPF to find a new path
when the primary path fails. With LFA FRR, OSPF pre-computes a
backup path and installs the backup next hop in the forwarding table.
Methods to calculate LFAs
Per-link: all prefixes that are reachable through a certain link all share
the same next hop address.
• An IGP can calculate a backup next hop for all prefixes that use the
same link.
• When the link fails, all prefixes will automatically be assigned to use
the same backup next hop address.
• The advantage of per link LFA is that it requires fewer CPU cycles and
memory than per-prefix LFA.
• The downside, however, is that once the primary link fails, you
suddenly put a lot of burden on the backup link.
Methods to calculate LFAs
Per-prefix: the IGP calculates an LFA for each and every prefix. It
requires more CPU cycles and memory but it does offer better load
balancing. When a primary path fails, prefixes could use different
backup paths, spreading the traffic throughout the network.
 Multihomed Scenario
ISP 1 - CNT
Branch Site

R2
10.1.1.0 .1

Internet
R1 172.16.1.0
.1 ISP 2-TELCONET

R3

 Assume that R1 has a multihomed connection to the Internet through ISP1-

CNT and ISP2-TELCONET.
 Two equal cost default static routes on R1 enable the to load balance
over the two links on a per-destination basis.
• R1 can detect if there is a direct failure on the link to one ISP, and in that case
use the other ISP for all traffic.
 Multihomed Scenario – Problem !!!!
ISP 1 - CNT
Branch Site

R2
10.1.1.0 .1

Internet
R1 172.16.1.0
.1 ISP 2-TELCONET

R3

 However, what would happen if a link within the ISP CNT provider
infrastructure were to fail?
• The link from R1 to R2 would still remain up and the R1 would continue to use that
link because the static default route would still be valid.
 How can this situation be corrected?
• Dynamic routing between R1 and the ISP networks; not practical !!!!
 Multihomed Scenario
ISP 1 - CNT
Branch Site

R2
10.1.1.0 .1

Internet
R1 172.16.1.0
.1 ISP 2-TELCONET

R3

 Another solution is to use either static routes or PBR on R1, but make them
subject to reachability tests toward critical destinations, such as the DNS
servers within the ISP.
• If the DNS servers in one of the ISPs go down or are unreachable, the static
route toward that ISP would be removed.
 These reachability tests can be performed with CISCO IP SLAs, Juniper RPM,
HPE NQA, etc.
• IP SLA and others can be configured on R1 to probe the DNS servers
frequently.
 IP SLAs
 The information collected can measure:
 Network resource availability
 Response time
 One-way latency
 Jitter (interpacket delay variance)
 Packet loss
 Voice-quality scoring
 Application performance
 Server response time
HUAWEI Network Quality Analysis (NQA)
The Network Quality Analysis (NQA) technology measures network performance and
Collects statistics on the delay, jitter, and packet loss ratio.

NQA monitors network quality of service (QoS) in real time and locates and diagnoses
network faults.

33
JUNIPER Real-Time Perfomance Monitor (RPM)
Real-time performance monitoring (RPM) allows you to perform service-level monitoring.
When RPM is configured on a device, the device calculates network performance based
on packet response time, jitter, and packet loss.

These values are gathered by Hypertext Transfer Protocol (HTTP) GET requests,
Internet Control Message Protocol (ICMP) requests, and TCP and UDP requests,
depending on the configuration.

34
 IP SLAs Operations
There are two types of IP SLAs
operations:
 Those in which the target device is
not running the IP SLAs responder
component (such as a web server
or IP host).
 Mostly ICMP generated traffic.

 Those in which the target device is

running the IP SLAs responder
component (such as a Cisco
router).
 Measurement accuracy is improved
when the target is a responder.
 Additional statistics can be
gathered.
IP SLA Configuration - Example

36
IP SLA Configuration - Example

37
IP SLA Configuration - Example
You have to use the ip sla schedule command to start your operation. You can
schedule it but we will start our operation right now and let it run forever

38
IP SLA Configuration - Example

39
IP SLA Configuration - Example

40
IP SLA Configuration – UDP Jitter Operation

41
Static Routing with IP SLA

42
Static Routing with IP SLA
ping IP address 192.168.12.2 each second and when we don’t get a response
within 100 Ms we will believe it’s unreachable

43
Static Routing with IP SLA
First I’ll remove the old default route and replace it with one that says “track 1”.
I can’t connect IP SLA directly with the static route, I have to do this with object
tracking. This is how you combine object tracking with IP SLA

track object-number ip sla operation-number {state | reachability}

44
Example IP SLA
OSPF 10
Example IP SLA
Example IP SLA

4
Example IP SLA

6
Lab. PBR &IP SLA - Homework