RSTP VS layer 2 Load ballencing

jeremy8529

Guys, I don't know if maybe you can give me some insight on a question that has been bugging me for a while, but I would like to know when it would be better to use RSTP than to simply use a load balancing protocol that works at layer two. Is the problem that broadcast loops still occur with most load balancing technologies, or is it just plain nasty to set up?

Thanks guys!

phoeneous

jeremy8529 wrote: »

Guys, I don't know if maybe you can give me some insight on a question that has been bugging me for a while, but I would like to know when it would be better to use RSTP than to simply use a load balancing protocol that works at layer two. Is the problem that broadcast loops still occur with most load balancing technologies, or is it just plain nasty to set up?

Thanks guys!

Nasty to setup? What do you mean?

How many vlans are you working with?

jeremy8529

Let's say we are working with 3 Vlans, and what I meant by nasty to set-up was, exceedingly difficult or complicated. What I really want to know is how you would have redundant switches and load balancing functioning at the same time. I know that by default STPs tend to put a port into a listening state until it detects an error in one of the other links then it turns into the forwarding state. To me, it seems like it would be best to have redundancy and aggregated connections, but i don't really know how that would work.

networker050184

You should always use STP which can handle the L2 redundancy piece for you by blocking links to prevent loops. To provide load balancing, you will just provide different roots for the different VLANs to have different forwarding paths. You can use a L3 redundancy protocol like HSRP to provide gateway redundancy for your hosts. These work together, and you don't use one over the other.

phoeneous

jeremy8529 wrote: »

Let's say we are working with 3 Vlans, and what I meant by nasty to set-up was, exceedingly difficult or complicated. What I really want to know is how you would have redundant switches and load balancing functioning at the same time. I know that by default STPs tend to put a port into a listening state until it detects an error in one of the other links then it turns into the forwarding state. To me, it seems like it would be best to have redundancy and aggregated connections, but i don't really know how that would work.

Not difficult at all. You can force the election of pvstp to any switch. So lets say you have three switches and three vlans, switch 1 can be the root of vlan 1, switch 2 can be the root of vlan 2, and switch 3 can be the root of vlan 3. This way not all vlans are relying on just one switch to be the root. This is a simplified load balancing method and I highly encourage it.

Read this:

Understanding and Configuring Spanning Tree Protocol (STP) on Catalyst Switches - Cisco Systems


P.S. You spelled balancing wrong.

ConstantlyLearning

I'll put a lab together tomorrow incorporating all that was mentioned before and post up the configs and diagrams.

networker050184

ConstantlyLearning wrote: »

I'll put a lab together tomorrow incorporating all that was mentioned before and post up the configs and diagrams.

Some people get paid good money to do that and you are giving it away for free! Good man

jeremy8529

You are a good man! I think tommorow that I will upload a drawing of mine to see if I have the right idea or not. I'm still a tad confused about how to physcily implement load balancing and STP on layer two at the same time, but I'm sure I can figure out. Thanks for the help guys!

Turgon

jeremy8529 wrote: »

Let's say we are working with 3 Vlans, and what I meant by nasty to set-up was, exceedingly difficult or complicated. What I really want to know is how you would have redundant switches and load balancing functioning at the same time. I know that by default STPs tend to put a port into a listening state until it detects an error in one of the other links then it turns into the forwarding state. To me, it seems like it would be best to have redundancy and aggregated connections, but i don't really know how that would work.

Perhaps try and define what you *want* to see happen a little more succinctly. You mention load balancing and STP. What do mean exactly? What are you trying to accomplish through technology implementation. Port channel will give you aggregated connections between switches and has load balancing features. STP will treat as a single link. L3 load balancing is something else, although you can have L3 port channels. STP cost parameters can give you options to send VLAN x traffic over one link as opposed to another, best cost to root switch so you can evenly share the traffic there when you have multiple links to places but use PVST+/RSTP.

A lot of load balancing is really concerned with layer 3 with devices using algorithms to hand off inbound TCP connections to less loaded backoffice systems. Layer 2 load balancing of course exists. But what are your desired requirements?

jeremy8529

"STP cost parameters can give you options to send VLAN x traffic over one link as opposed to another, best cost to root switch so you can evenly share the traffic there when you have multiple links to places but use PVST+/RSTP."

That there might have answered my question. Thanks!

What I want to finally accomplish is this, I want to be able to have redundant switches and links and at the same time maximize network performance by utilizing the switches I have implemented for redundancy. Because I thought that the redundant switch ports were not used unless there was a failure when you used a STP. Put simply, I am wanting to address performance, scalability, and reliability the best I can on a budget on layer 2.

Cheers!

ConstantlyLearning

Here's the lab I've done up.

Disclaimer: I'm a novice at this too.

RSTP is configured on all switches.

There are 4 VLAN's in the switched network which were configured on Dist_SW_1 and propagated out using VTP.

I've lowered the STP priority on Dist_SW_1 to 4096 for VLAN's 1 and 2 making it the root bridge for those VLAN's. I lowered the STP priority on the same switch for VLAN's 3 and 4 to 8192 making it the backup root bridge for those VLAN's. The reverse is configured on Dist_SW_2.
This creates load sharing of the BPDU's, instead of all being sent and received by a single root bridge they are spilt across two root bridges. There is also root bridge redundancy

HSRP is configured on Dist_SW_1 and Dist_SW_2 to provide default gateway redundancy. I've configured Dist_SW_1 to be the active gateway for VLAN's 1 and 2 and the standby gateway for VLAN's 3 and 4. The reverse is configured on Dist_SW_2.
If an end device is connected to a switch port on VLAN 4, you would assign it a default gateway of the the virtual IP of the HSRP group for VLAN 4. Any traffic destined for an unknown network would be sent to Dist_SW_2 as that is the active HSRP gateway for VLAN 4. If Dist_SW_2 dies or the link from an access switch to Dist_SW_2 dies then Dist_SW_1 would become the active HSRP gateway.
So, this provides both redundancy and load sharing.




Output of 'show spanning-tree' command on Dist_SW_1 (Some details removed for brevity)

Dist_SW_1#sh spanning-tree

VLAN0001
Spanning tree enabled protocol ieee
Root ID Priority 4097
Address 0013.1ad3.f800
This bridge is the root

Bridge ID Priority 4097 (priority 4096 sys-id-ext 1)
Address 0013.1ad3.f800


VLAN0002
Spanning tree enabled protocol ieee
Root ID Priority 4098
Address 0013.1ad3.f800
This bridge is the root

Bridge ID Priority 4098 (priority 4096 sys-id-ext 2)
Address 0013.1ad3.f800


VLAN0003
Spanning tree enabled protocol ieee
Root ID Priority 4099
Address 0012.d982.5680

Bridge ID Priority 8195 (priority 8192 sys-id-ext 3)
Address 0013.1ad3.f800


VLAN0004
Spanning tree enabled protocol ieee
Root ID Priority 4100
Address 0012.d982.5680

Bridge ID Priority 8196 (priority 8192 sys-id-ext 4)
Address 0013.1ad3.f800


Output of 'show spanning-tree' command on Dist_SW_2 (Some details removed for brevity)


VLAN0001
Spanning tree enabled protocol ieee
Root ID Priority 4097
Address 0013.1ad3.f800


Bridge ID Priority 8193 (priority 8192 sys-id-ext 1)
Address 0012.d982.5680



VLAN0002
Spanning tree enabled protocol ieee
Root ID Priority 4098
Address 0013.1ad3.f800

Bridge ID Priority 8194 (priority 8192 sys-id-ext 2)
Address 0012.d982.5680



VLAN0003
Spanning tree enabled protocol ieee
Root ID Priority 4099
Address 0012.d982.5680
This bridge is the root

Bridge ID Priority 4099 (priority 4096 sys-id-ext 3)
Address 0012.d982.5680



VLAN0004
Spanning tree enabled protocol ieee
Root ID Priority 4100
Address 0012.d982.5680
This bridge is the root

Bridge ID Priority 4100 (priority 4096 sys-id-ext 4)
Address 0012.d982.5680


Output for 'show standby' command on Dist_SW_1

Vlan1 - Group 1
Local state is Active, priority 150, may preempt
Hellotime 100 msec, holdtime 400 msec
Next hello sent in 0.092
Virtual IP address is 10.0.0.10 configured
Active router is local
Standby router is 10.0.0.2 expires in 0.292
Virtual mac address is 0000.0c07.ac01
2 state changes, last state change 06:56:34
IP redundancy name is "hsrp-Vl1-1" (default)
Vlan2 - Group 2
Local state is Active, priority 150, may preempt
Hellotime 100 msec, holdtime 400 msec
Next hello sent in 0.022
Virtual IP address is 10.0.1.10 configured
Active router is local
Standby router is 10.0.1.2 expires in 0.292
Virtual mac address is 0000.0c07.ac02
2 state changes, last state change 06:44:26
IP redundancy name is "hsrp-Vl2-2" (default)
Vlan3 - Group 3
Local state is Standby, priority 100, may preempt
Hellotime 100 msec, holdtime 400 msec
Next hello sent in 0.065
Virtual IP address is 10.0.2.10 configured
Active router is 10.0.2.2, priority 150 expires in 0.352
Standby router is local
4 state changes, last state change 06:42:01
IP redundancy name is "hsrp-Vl3-3" (default)
Vlan4 - Group 4
Local state is Standby, priority 100, may preempt
Hellotime 100 msec, holdtime 400 msec
Next hello sent in 0.049
Virtual IP address is 10.0.3.10 configured
Active router is 10.0.3.2, priority 150 expires in 0.356
Standby router is local
4 state changes, last state change 06:41:39
IP redundancy name is "hsrp-Vl4-4" (default)


Output for 'show standby' command on Dist_SW_2

Vlan1 - Group 1
State is Standby
1297 state changes, last state change 06:33:41
Virtual IP address is 10.0.0.10
Active virtual MAC address is 0000.0c07.ac01
Local virtual MAC address is 0000.0c07.ac01 (v1 default)
Hello time 100 msec, hold time 400 msec
Next hello sent in 0.092 secs
Preemption enabled
Active router is 10.0.0.1, priority 150 (expires in 0.384 sec)
Standby router is local
Priority 100 (default 100)
IP redundancy name is "hsrp-Vl1-1" (default)
Vlan2 - Group 2
State is Standby
2494 state changes, last state change 06:32:11
Virtual IP address is 10.0.1.10
Active virtual MAC address is 0000.0c07.ac02
Local virtual MAC address is 0000.0c07.ac02 (v1 default)
Hello time 100 msec, hold time 400 msec
Next hello sent in 0.068 secs
Preemption enabled
Active router is 10.0.1.1, priority 150 (expires in 0.296 sec)
Standby router is local
Priority 100 (default 100)
IP redundancy name is "hsrp-Vl2-2" (default)
Vlan3 - Group 3
State is Active
1 state change, last state change 06:41:00
Virtual IP address is 10.0.2.10
Active virtual MAC address is 0000.0c07.ac03
Local virtual MAC address is 0000.0c07.ac03 (v1 default)
Hello time 100 msec, hold time 400 msec
Next hello sent in 0.088 secs
Preemption enabled
Active router is local
Standby router is 10.0.2.1, priority 100 (expires in 0.372 sec)
Priority 150 (configured 150)
IP redundancy name is "hsrp-Vl3-3" (default)
Vlan4 - Group 4
State is Active
1 state change, last state change 06:40:37
Virtual IP address is 10.0.3.10
Active virtual MAC address is 0000.0c07.ac04
Local virtual MAC address is 0000.0c07.ac04 (v1 default)
Hello time 100 msec, hold time 400 msec
Next hello sent in 0.084 secs
Preemption enabled
Active router is local
Standby router is 10.0.3.1, priority 100 (expires in 0.336 sec)
Priority 150 (configured 150)
IP redundancy name is "hsrp-Vl4-4" (default)


You could also have aggregated links between the switches using etherchannel.


Anyone got anything to add?

jeremy8529

Oh, so you could have aggregated trunk lines is what you are saying. So when you physcily wired this baby up, you would have two fast ethernet cables running from access switch one to dist switch one for the root connection, and then the same for the redundent connection correct? This would make the actual fast ethernet cables redudent and also share the load between trunk ports right?

ConstantlyLearning

jeremy8529 wrote: »

Oh, so you could have aggregated trunk lines is what you are saying. So when you physcily wired this baby up, you would have two fast ethernet cables running from access switch one to dist switch one for the root connection, and then the same for the redundent connection correct? This would make the actual fast ethernet cables redudent and also share the load between trunk ports right?

Yes, a trunk between two switches could be made up of aggregated links using etherchannel. This would provide a greater amount of bandwidth between the switches and the logical link would remain up if any of the physical links go down. (As long as there is still one physical link up in the etherchannel group).

I think you're viewing the distribution switches just as 'root' and 'backup root' when you should be viewing them as root and backup root "for certain VLAN's".

jeremy8529

[QUOTE=ConstantlyLearning;388936
I think you're viewing the distribution switches just as 'root' and 'backup root' when you should be viewing them as root and backup root "for certain VLAN's".[/QUOTE]

Aye, I think I know what you are talking about now. The back up is only a back up from the perspective on a certain VLAN, and to another it might be the primary.

ConstantlyLearning

jeremy8529 wrote: »

Aye, I think I know what you are talking about now. The back up is only a back up from the perspective on a certain VLAN, and to another it might be the primary.

Yeah, exactly.

Is there a specific switched network already in place that you are looking to improve or are you "just wondering how all this redundancy and efficiency stuff works"?

Post up an image of the switched network you have if you can.

phoeneous

jeremy8529 wrote: »

Oh, so you could have aggregated trunk lines is what you are saying. So when you physcily wired this baby up, you would have two fast ethernet cables running from access switch one to dist switch one for the root connection, and then the same for the redundent connection correct? This would make the actual fast ethernet cables redudent and also share the load between trunk ports right?

You can even have up to 8 links for a single EtherChannel.

jeremy8529

I am mainly trying to figure this stuff out for a network design competition that I am participating in that is targeted at high school students. Plus considering that I plan on majoring in computer engineering, knowledge is power and by god eventually I want to become a "sith lord" of computer know how. It also doesn't help that this stuff is pretty cool!

@phoeneous
Wow, I am trying to figure out who exactly would need that!

phoeneous

jeremy8529 wrote: »

I am mainly trying to figure this stuff out for a network design competition that I am participating in that is targeted at high school students. Plus considering that I plan on majoring in computer engineering, knowledge is power and by god eventually I want to become a "sith lord" of computer know how. It also doesn't help that this stuff is pretty cool!

@phoeneous
Wow, I am trying to figure out who exactly would need that!

Anyone who wants to take advantage of redundancy and has enough ports to do so. Look up 'core, access, and distribution'.