Advertenties

ProCurve Network Redundancy

If you run a network that requires redundant links then spanning-tree is a feature you are going to need. With the default spanning-tree configuration enabled you will have a failover time of a few seconds opening a new path.

Broadcast Storms
In any normal situation it is illegal to wire Ethernet in a loop, when this is done a broadcast storm will occur since all packages are ‘looped’ throughout the network. Spanning-tree is a protocol designed to detect/monitor multiple paths to the same destination and disable one or more of these paths if that seems necessary.

Redundancy
That being said, the spanning-tree feature is useful when creating a fault tolerant, and redundant, network. Spanning-tree can restore broken connections/paths by enabling (or reactivating) another connection.

One of the best options is to configure 2 switches as ‘core’ switches within your network, all other switches will be connected to both of these core switches and are called ‘edge’ switches. This means every switch has 2 paths (core1/core2) to reach it’s destination. Spanning-tree will prevent path 2 from being activated until path 1 has failed.

Example configuration:
On the core switches you can enable spanning-tree with the following commands:
(config) # spanning-tree config-name “C1”
(config) # spanning-tree config-revision 1
(config) # spanning-tree instance 1 vlan 1-299
(config) # spanning-tree instance 2 vlan 299-350
(config) # spanning-tree

Now you have setup and enabled spanning-tree for vlan 1 until 350. You can now set the priority of the core switch (in my case core switch 1).
(config) # spanning-tree instance 1 priority 0
(config) # spanning-tree instance 2 priority 1
(config) # spanning-tree priority 0

These commands tell the switch that Core 1 is the main switch for instance 1 and acts as secondary for instance 2. On the second core switch the config is almost identical except for the last few commands (the priority part):
(config) # spanning-tree instance 1 priority 1
(config) # spanning-tree instance 2 priority 0
(config) # spanning-tree priority 0

Configuration of the edge switches is similar to the ones we used above;
(config) # spanning-tree config-name “C1”
(config) # spanning-tree config-revision 1
(config) # spanning-tree instance 1 vlan 1-299
(config) # spanning-tree instance 2 vlan 299-350
(config) # spanning-tree

After you have made trunks from one edge switch to both core switches and enabled them you should be able to see spanning-tree in action. Log on to the edge switch using telnet and issue the following command:
sh spanning-tree PORT instance 1

The port would be one of the trunks you have configured to see what spanning-tree has chosen to do with it. The output will show you ports that are set to ‘Forwarding’ – meaning it’s active – and ‘Blocking’ as secondary path.

If you want to see all ports and there spanning-tree configuration at that time just issue:
sh spanning-tree

To see what switch is connected to that specific port you can use LLDP. Issue the following command:

sh lldp info remote

You will get a list of all ports and the connected device, this way you can verify if the port is connected to the switch you want without being on site.

If a cable between any two switches is cut, or a switch fails, disabled paths are automatically reactivated. This means that when a switch has crashed, only the hosts that are directly connected to this switch will suffer a network failure and the rest of the network stays online. A cable that gets cut should have no influence since it will activate the other path directly.

Failover test
When performing a test in our network we check what would happen if we experience a uplink failure (cable broken?) of a core switch failure. Failover time was less then 3 seconds in both scenario’s. The uplink failure can also be prevented by using a trunk of several ports to act as one uplink.

Advertenties

Geef een reactie

%d bloggers liken dit: