Posts
BGP Notes
BGP Messages and Neighbor states
- idle
- connect
- active
- open sent
- open confirm
- established
- open - neighbor relationship/exchange, basic parameters
- keepalive- to maintain neighbor relationship
- Update - exchange routing info
- notification - bgp error/ reset neighbor relationship
keepalive hold time 60-180 sec
neighbor update-source ->> when source IP address is used to reach destination.
neighbor command --> when destination address is used
auto-summary - off by default
authentication - MD 5 only
BGP update message format ->
2 bytes
length of the withdrawn routes section
withdrawn routes (variable)
length of the path attribute section
path attributes (variable)
prefix length| prefix
prefix length| prefix
................................
IP Protocol Number
- ICMP - 1
- TCP - 6
- UDP - 17
- OSPF - 89
- EIGRP - 88
- PIM - 103
- IGMP - 2
OSPF Notes ..
Wednesday, July 16, 2008
5 different types of msgs TO EXCHANGE LSAs.
LSA (Link State Advertisement) is NOT A Message.
LSA is a data structure which is held inside routers' LSDB and exchanged using LSU
5 messages --> hello - To discover each others' presence/ monitor the neighbor for any change in topology
-->Link State Database Description (who's missing..LSDBD) -->
--> LSR -link state request --> request for an LSA
---> LSU -link state update --> exchange an LSA
---> LSAck --Link State Acknowledgment --> ack the receipt of an LSA
Neighbor state FROM NULL TO FULL>>>
RID 1.1.1.1--------------------------------------RID 2.2.2.2
------------>Hello (Null) RID 1.1.1.1----------->
<------------Hello ( 1.1.1.1) RID 2.2.2.2<---------------- -------------->Hello (1.1.1.1, 2.2.2.2) RID 1.1.1.1-------->
<--------------hello DR z.z.z.z------------->
--------------> DD LSA headers------------------>
<----------------DD LSA headers------------>
LOADING STATE <---- LSR, LSU, LSAck-----> LOADING STATE
FULL -----------------------------------FULL
- down
- init
- 2-way
- ex-start
- exchange
- loading
- full
all this is stored in STATE MACHINES for each neighbor in each router.
primary address - for use to send hellos
secondary address for advertising..
to pass a successful hello -- else no neighbor relationship
---> authentication pass type0, type 1, type MD5
--->same subnet & subnet mask
---> same area
---> same area type
--->no duplicate RIDs
---> hello and death timers must be equal
others.
PID shouldnt be the same
MTU for the DD packets must be equal.
hello timer - 10 sec for LAN
30 sec for WAN T1 slow links
death timer - 4 times the hello timer
How to send LSAs?
--> flooding..
--> request is required.
- One DD packet can have multiple LSAs inside it. (headers only)
- each DD packet has a sequence number
- ack by sending the identical DD packet
- wait for ack before sending a new DD packet
- new LSA starts from 0x80000001 back to 0x7fffffff
- LSU has full LSA
DR and BDRs--->
DR would be flooding and creating LSA type2.
How to select DR and BDRs?
all otther routers including BDR--------> send hello using 224.0.0.6---> DR
<-----------unicast ack by DR to all of them---<---- <-------same packet using 224.0.0.5 to all the routers---<> once hello is exchanged.
when adjacency is formed -----> full data base axchange
hello packet has DR field. if sent empty..doesnt want to become a DR
send hello with priority 125 in it.
highest priority if tie then heightest RID is choosen a DR
the one who doesnt want to become a DR is made a BDR with same characteristic.
routers dont adv/claim themself once they know the other is more suitable for the position.
DR fails --> BDR becomes the DR
OSPF Network type on the basis of using or not using DR/BDR, default hello, 2 host on the subnet etc.
--> Broadcast
-->Point to point
---> NBMA -->non broadcast multiaccess
--> Point to multipoint
--> point to multipoint non broadcast.
--> loopback
LSA Types
type1 - router >>one per router for all the interface IP addresses/stub networks
type2- network >>one per transit network created by DR for subnetting
type3- Net summery >> created by ABR for area type like for going in different arrea to carry orgin area plus cost but not to carry topology.
type4 - ASBR Summery >> like 3 for a hostroute to reach ASBR
type5- AS external >>by ASBR for external routes.
the rest all somehow not used much like:
type6 - group membership >> for MOSPF not on cisco IOS
type 7- NSSA external >> for NSSA areas used as type 5 (NSSA is not so stubby area)
type8- external attribute - not on cisco routers
type 9-11 for future. but type10 is used in MPLS traffic engineering.
Steady state:
hello pass with in the death interval.
per LSA link state refresh advertised by default with in 30 minutes.
and refreshed with in 60minutes default max age timer.
GOALS.--> reduce convergence time
--> reduce overhead processing --use a large number of small small areas.
--> improve network stability if link flaps happens then in only that area calculation happens.
ABR. ASBR to do route summerizations so less number of LSA type3
Note: type3 is summary LSA
summary means summary-address and area-range
this refers to LSA headers
Stubby area: some rules applied in that particular area like using default routes,
only one subnet etc.
route filtering using distributed list -- dont want some routes just filter them out.
this is no filtering LSA.
then filtering LSAs like not using LSA3
area range also comes in here somewhere.,
virtual link -- not directly connected to area 0 connect with the ABR of the area which is directly connected to the area0
what an general LSA is having inside? lsiting of routes with their cost.
show ip ospf database summary
packet format also : in bytes 24 bytes.
version -1| packet type 1| packet length 2| RID 4| areaID 4 | checksum 2| authenticationtype 2| authentication 8| data variable|
additional notes later .....
Read more...
RIP Notes..
Split horizon -- poison reverse--
instead of advertising all the routes, RIP omits out those routes from the table who's outgoing interface field matches to the update interface.
loop prevention: - which fails --> hop count /metric 16
QoS Notes..
ISL - cos 3 bits
802.1q/p - 3 bits
dest | src | ethernet type | tag | --------------
|
|
| | | | | vlan id |
user
priority
2 bytes
FR --> DE --> Discard Eligibility 1 bit
ATM --> CLP --> Cell Loss Priority 1bit
MPLS --> EXP --> 3 bits
only for classification n marking --> QoS tools on edge devices.
CB -> class-based --> marking
--> policing
--> shaping
--> header compression
--> WFQ --> CBWFQ
marking: first mark the packet and then set the appropriate QoS feature to marked traffic.
marking (solely) is used to identify the packets (exception WRED/DWRED -use marking to detect and drop packets)?
like CLP 1 (ATM) and DE 1 (FR) packets are dropped by wred and dwred before the ones with CLP and DE bits 0 when congestion happens.
marking is supported on int, subint, ATM pvcs.
it can be configured in the same policy in which queueing actions are configured.
marking is done by
-IP Packets precedence (8 bits) dscp (64 bits)
-setting cos for layer2
-associating local qos group value (up to 100 diff values ranging from community string, prefix, AS, but since its local to the router, cant be used in output interface policy)
-settin CLP for ATM from 0 to 1
-setting DE for FR from 0 to 1
IGMP Notes...
Monday, July 14, 2008
Internet Group Management Protocol :-
IGMP message --> IP Datagram sent with protocol number 2 and TTL 1.
Layer-2 ( RGMP - route-port
CGMP - Cisco
IGMP Snooping.)
0x11 query
0x12 report v1
0x16 report v2
0x22 report v3
0x17 leave group v2
01005E.0.00001 - 224.0.0.1
Query to 224.0.0.1 to hosts
reply to 224.0.0.2 to routers
Solicited Reports --> On Receiving a Query
Unsolicited Report --> On Joining Group
Report Suppression
MRT (Max Response time) - v1 -10 sec
Leave latency - 60 secs v1
- immediate - v2 (3 minutes to 3 seconds)
Interface address - source address
the route with the lowest IP address
PIM - 224.0.0.13
OSPF 5/6
Snooping: first look at the routing protocol message like hello message.
Scoping: TTL scoping (Interface and packet TTL)/Administrative Scoping ( use of 239.0.0.0-255.255.255)
Subscribe to:
Posts (Atom)
