- Posts: 350
- Thank you received: 0
Spanning Tree Protocol
18 years 3 months ago #16473
by jwj
-Jeremy-
Replied by jwj on topic Re: Spanning Tree Protocol
Start with the original spanning tree, 802.1d, and continue from there. That forms the basis for understanding spanning tree.
-Jeremy-
18 years 3 months ago #16579
by koslyr
Replied by koslyr on topic Re: Spanning Tree Protocol
Some basics about STP.
One main question is why we need Spanning Tree Protocol? We need the STP (802.1D) because is very possible to exist physical loops (redundancy path as a part of high reliability system) in our network, which leads some switches recognize the same stations on more than one side of the switch (really hell to MAC tables – broadcast storms).This is when a single broadcast frame (eg. data packets) will circulate endlessly and consumes all available bandwidth on the loop making the network unusable. The redundancy path (2 paths between 2 nodes) in a network environment is a very common strategy in order to make sure that data transmission won't be interrupted due to link failures. When we set up the network however, packets will probably be sent more than twice if a switch finds multiple paths that can be chosen. This is where the situation gets serious causing packet flooding, and greatly decreases the network bandwidth. The solution (activation of STP) is to block one segment of the network and the blocked path will be enabled only when the primary one fails, providing a backup path for data transmission. So STP is a protocol for network redundancy which creates a loop free logical topology. If we have a network with redundancy paths for protection is vital to make activate the STP for forcing certain redundant data paths into standby (blocked) state. More specific the STP assigns roles (root, designated, blocking) to each port of the switches depending of the BPDU messages.
So if one link becomes failed (or a switch fails), the nearest switch produces a Topology Change BPDU (TC BPDU) in order the STP algorithm reconfigures the Spanning Tree topology and activate the redundant backup links. The real problem with STP is the too long recovery time for activating the standby data path (30-60 sec in a large network). When a link failure occurs, the switches over that network often spend much time on exchanging STP messages and synchronization. In this case the blocking port is opened (transition port state to forwarding), thus re-establishing a path for the network traffic. But the STP performance in terms of recovery time is too slow for many real time applications. To avoid this problem, we are using the RSTP (Radid STP, 802.1w) which simplifies the information that is being communicated and increases the recovery speed (3-6 sec) of a network backup path. The RSTP shares the same basic concepts with STP (fully compatible) but the main advantage is that offers a fast network redundancy solution. Here we must mention that these recovery times are far away from the SDH restoration time (50 msec). This is another big topic about Metro Ethernet and the choice between Carrier Ethernet (pure Ethernet) or Ethernet Over NG-SDH at the MAN networks.
Expect of STP and RSTP there is also the MSTP (Multiple STP, 802.1s) which is the perfect combination if we are using in our networks the VLAN load balance. This means that the MSTP runs a separate instance of RSTP for each individual VLAN configured in the network. I am not going to expand any more because I have written too much for a post.
For closing I pose some questions:
1)At the layer 3 (routers) there is also the problem of redundancy data paths. How the routers address the packet flooding because the STP is only for Layer 2 (Ethernet)? This has to do with the field of TTL which resides at IP packets?
2)What is the difference between the Per Vlan Spanning Tree Protocol (PVST) and Multiple Spanning Tree Protocol (MSTP)?
One main question is why we need Spanning Tree Protocol? We need the STP (802.1D) because is very possible to exist physical loops (redundancy path as a part of high reliability system) in our network, which leads some switches recognize the same stations on more than one side of the switch (really hell to MAC tables – broadcast storms).This is when a single broadcast frame (eg. data packets) will circulate endlessly and consumes all available bandwidth on the loop making the network unusable. The redundancy path (2 paths between 2 nodes) in a network environment is a very common strategy in order to make sure that data transmission won't be interrupted due to link failures. When we set up the network however, packets will probably be sent more than twice if a switch finds multiple paths that can be chosen. This is where the situation gets serious causing packet flooding, and greatly decreases the network bandwidth. The solution (activation of STP) is to block one segment of the network and the blocked path will be enabled only when the primary one fails, providing a backup path for data transmission. So STP is a protocol for network redundancy which creates a loop free logical topology. If we have a network with redundancy paths for protection is vital to make activate the STP for forcing certain redundant data paths into standby (blocked) state. More specific the STP assigns roles (root, designated, blocking) to each port of the switches depending of the BPDU messages.
So if one link becomes failed (or a switch fails), the nearest switch produces a Topology Change BPDU (TC BPDU) in order the STP algorithm reconfigures the Spanning Tree topology and activate the redundant backup links. The real problem with STP is the too long recovery time for activating the standby data path (30-60 sec in a large network). When a link failure occurs, the switches over that network often spend much time on exchanging STP messages and synchronization. In this case the blocking port is opened (transition port state to forwarding), thus re-establishing a path for the network traffic. But the STP performance in terms of recovery time is too slow for many real time applications. To avoid this problem, we are using the RSTP (Radid STP, 802.1w) which simplifies the information that is being communicated and increases the recovery speed (3-6 sec) of a network backup path. The RSTP shares the same basic concepts with STP (fully compatible) but the main advantage is that offers a fast network redundancy solution. Here we must mention that these recovery times are far away from the SDH restoration time (50 msec). This is another big topic about Metro Ethernet and the choice between Carrier Ethernet (pure Ethernet) or Ethernet Over NG-SDH at the MAN networks.
Expect of STP and RSTP there is also the MSTP (Multiple STP, 802.1s) which is the perfect combination if we are using in our networks the VLAN load balance. This means that the MSTP runs a separate instance of RSTP for each individual VLAN configured in the network. I am not going to expand any more because I have written too much for a post.
For closing I pose some questions:
1)At the layer 3 (routers) there is also the problem of redundancy data paths. How the routers address the packet flooding because the STP is only for Layer 2 (Ethernet)? This has to do with the field of TTL which resides at IP packets?
2)What is the difference between the Per Vlan Spanning Tree Protocol (PVST) and Multiple Spanning Tree Protocol (MSTP)?
18 years 3 months ago #16591
by jwj
-Jeremy-
Replied by jwj on topic Re: Spanning Tree Protocol
I'm not sure what exactly you mean with #1, but the difference between PVST and MST is that PVST is Cisco proprietary. It's like the difference between PAGP (Cisco only) and LAGP (IEEE standard) for etherchannels. There are some differences in their terminology, but the main difference between PVST and MST is that MST is meant mostly to group VLANs into spanning tree instances. Each instance of spanning tree uses more CPU, so MST can offer VLAN balancing without potentially bringing your switch to it's knees.
-Jeremy-
Time to create page: 0.125 seconds