STP help

[SANTHOSHKUMAR SARAVANAN]

HI ALL ,

              I have 2 user distribution switch (UD1, UD2) , 6 user access of 3550 (Sw1,Sw2,Sw3,Sw4,Sw5,Sw6) , my total End user count is 280 nos , connectvity between two UD switch is L3 links ,All user access switch uplinks are connected to two UD block (UD1 ,UD2)  . I have SVI l3 VLAN 192.168.1.0/23 on both UD switches with HSRP running for gateway redudancy

Ud1: 192.168.1.3

ud2: 192.168.1.4

VIP : 192.168.1.1

primary STP root bridge is UD1, Secondary root bridge is UD2 . In normal case all user access switches,  root port will be the interface connecting to  UD1 , among all switches some two port will be in blocked state .

               For eg in Sw2 interface connecting to ud1 had gone to error disable mode and line protocol was down. 

                 Now traffic from Sw2 pointing to default gateway 192.168.1.1 follows as such

(Sw2-> Ud2-> Sw1- Ud1 ) or it take best cost to root bridge . so it has got additional burden to Sw1 to process Sw2 traffic also .

                    which would be best solution on this scenario

1) L2 access link between UD block

2) GLBP for gateway redudancy without using l2 access link ..

[Jon Marshall]

sanvaishu wrote:
HI ALL ,
              I have 2 user distribution switch (UD1, UD2) , 6 user access of 3550 (Sw1,Sw2,Sw3,Sw4,Sw5,Sw6) , my total End user count is 280 nos , connectvity between two UD switch is L3 links ,All user access switch uplinks are connected to two UD block (UD1 ,UD2)  . I have SVI l3 VLAN 192.168.1.0/23 on both UD switches with HSRP running for gateway redudancy 
Ud1: 192.168.1.3
ud2: 192.168.1.4
VIP : 192.168.1.1
primary STP root bridge is UD1, Secondary root bridge is UD2 . In normal case all user access switches,  root port will be the interface connecting to  UD1 , among all switches some two port will be in blocked state . 
               For eg in Sw2 interface connecting to ud1 had gone to error disable mode and line protocol was down.  
                 Now traffic from Sw2 pointing to default gateway 192.168.1.1 follows as such 
(Sw2-> Ud2-> Sw1- Ud1 ) or it take best cost to root bridge . so it has got additional burden to Sw1 to process Sw2 traffic also . 
                    which would be best solution on this scenario
1) L2 access link between UD block 
2) GLBP for gateway redudancy without using l2 access link ..

If your interconnect between UD1 and UD2 is a L3 link then there should be no blocking from the access-layer switches because there should be no L2 loop. Just to confirm -

1) an access-layer switch eg. sw1 is connected to both UD1 & UD2 ?

2) the links are either single links to UD1 and UD2 or they are etherchannel links ?

3) sw1 does not have any direct connections to any of the other sw switches ?

Edit - when you say the link between UD1 & UD2 is L3 can you post the config of the link ?

Jon

[SANTHOSHKUMAR SARAVANAN]

Hi John,

               I have same vlan 10 for all User access switches as i said my total user count is 280 nos

1) an access-layer switch eg. sw1 is connected to both UD1 & UD2 ? yes all access layer switch (sw1, sw2,sw3,sw4,sw5,sw6) is connected to UD1 & UD2

2) the links are either single links to UD1 and UD2 or they are etherchannel links ? single link from all switch gi1/1 connect to Ud1 & gi1/2 connect to UD2

3) sw1 does not have any direct connections to any of the other sw switches ? : No direct connection to user access switches

Edit - when you say the link between UD1 & UD2 is L3 can you post the config of the link ? : I am running OSFP between UD1, UD2 , router ,

UD1

int gig 1/2

no switchport

ip address 10.150.100.9 255.255.255.252

router osf 1

network 10.150.100.8 0.0.0.3 area 0

network 10.150.100.4 0.0.0.3 area 0

network 192.168.1.0  0.0.1.255 area 0

passive interface default

no passive interface int gig 1/1

no passive interface int gig 1/2

UD2

int gig 1/2

no switchport

ip address 10.150.100.10 255.255.255.252

router osf 1

network 10.150.100.8 0.0.0.3 area 0

network 10.150.100.0 0.0.0.3 area 0

network 192.168.1.0  0.0.1.255 area 0

passive interface default

no passive interface int gig 1/1

no passive interface int gig 1/2

[Jon Marshall]

Ahh, okay, it's because you have the same vlan on all access switches. Personally i would do the following -

1) each access-switch should have it's own vlan

2) UD1 & UD2 should then route between the vlans

3) use GLBP on UD1 & UD2

if you did the above then each access-layer switch could use both uplinks for forwarding and would spread the load between the uplinks due to GLBP.

Edit -making the link between UD1 & UD2 would not help at all as this only adds a further L2 loop into your topology. If you are uncomfortable using GLBP you could still use HSRP and have UD1 active for the vlans on sw1,3.5 and UD2 active for sw2,4,6.

Jon

[SANTHOSHKUMAR SARAVANAN]

Hi John ,

               Could u please suggest me which would be best solution ,

1) If HSRP i  use,  we need to go with one new L2 Link btwn both UD1 , UD2  along with L3 Link for routing . To pass the traffic directly from UD2  to  UD1 . If any uplink links connecting to UD1  fails on user access switches .

2) If i use GLBP , I dont want have additional L2 link .  UD2 switch will automatically take care of forwarding traffic . because it would one of AVF for a AVG .

which could suit more , HSRP or GLBP ???

[Jon Marshall]

sanvaishu wrote:
Hi John ,
               Could u please suggest me which would be best solution ,
1) If HSRP i  use,  we need to go with one new L2 Link btwn both UD1 , UD2  along with L3 Link for routing . To pass the traffic directly from UD2  to  UD1 . If any uplink links connecting to UD1  fails on user access switches .
2) If i use GLBP , I dont want have additional L2 link .  UD2 switch will automatically take care of forwarding traffic . because it would one of AVF for a AVG .
which could suit more , HSRP or GLBP ???

You need to reread my other answers. With your current setup doesn't really matter what you do because you have the same vlan across all switches.

1) You don't need a L2 link between UD1 & UD2 if you want to run HSRP although it could help in traffic paths - see below

2) It depends on which link is blocked. For example if sw1 is forwarding to UD1 and blocking to UD2 and UD1 hands out UD2 then somehow traffic needs to get to UD2. It would have to go sw1 -> UD1 -> another access switch to UD2 because it cannot go across the L3 link.

So with either HSRP or GLBP with this setup you have very suboptimal paths. Your problem is the same vlan on all access switches.

Jon