Switching

Hi People,

Currently, I have four Cores separate for my internal network. Two ex4600 for Users (End-user, WiFi, Telephony, Print, Cameras, etc) and the other two Ex4600 for Servers.
The Users' Core (It has configure BGP, VRRP, DHCP Relay, VLAN L2/L3, etc) and The Server's Core (It has configure BGP, VRRP, etc). All Cores are on going.

My question is: Can I configure MC-LAG on all ex4600?, MC-LAG is compatibility with all those feature? (BGP, VRRP, DHCP Relay, etc)

I understand that it will be affect the network, but not problem. 

Tks for your help!

Hi,

EX4600 not support MC-LAG. You can check at juniper pathfinder.


Thanks

Hi People,

Currently, I have four Cores separate for my internal network. Two ex4600 for Users (End-user, WiFi, Telephony, Print, Cameras, etc) and the other two Ex4600 for Servers.
The Users' Core (It has configure BGP, VRRP, DHCP Relay, VLAN L2/L3, etc) and The Server's Core (It has configure BGP, VRRP, etc). All Cores are on going.

My question is: Can I configure MC-LAG on all ex4600?, MC-LAG is compatibility with all those feature? (BGP, VRRP, DHCP Relay, etc)

I understand that it will be affect the network, but not problem. 

Tks for your help!

Hi People,

Currently, I have four Cores separate for my internal network. Two ex4600 for Users (End-user, WiFi, Telephony, Print, Cameras, etc) and the other two Ex4600 for Servers.
The Users' Core (It has configure BGP, VRRP, DHCP Relay, VLAN L2/L3, etc) and The Server's Core (It has configure BGP, VRRP, etc). All Cores are on going.

My question is: Can I configure MC-LAG on all ex4600?, MC-LAG is compatibility with all those feature? (BGP, VRRP, DHCP Relay, etc)

I understand that it will be affect the network, but not problem. 

Tks for your help!

Hello 

I don't really understand what MC-LAG has to do with bgp, dhcp etc.
You pin the EX4600 in virtual chassie and then you can use MC-LAG 

What and how you do on this lag is another matter 

Hi People,

Currently, I have four Cores separate for my internal network. Two ex4600 for Users (End-user, WiFi, Telephony, Print, Cameras, etc) and the other two Ex4600 for Servers.
The Users' Core (It has configure BGP, VRRP, DHCP Relay, VLAN L2/L3, etc) and The Server's Core (It has configure BGP, VRRP, etc). All Cores are on going.

My question is: Can I configure MC-LAG on all ex4600?, MC-LAG is compatibility with all those feature? (BGP, VRRP, DHCP Relay, etc)

I understand that it will be affect the network, but not problem. 

Tks for your help!

Hi GRZEGORZ DACKA

I get it, ok tks!

Hello 

I don't really understand what MC-LAG has to do with bgp, dhcp etc.
You pin the EX4600 in virtual chassie and then you can use MC-LAG 

What and how you do on this lag is another matter 

Hi People,

Currently, I have four Cores separate for my internal network. Two ex4600 for Users (End-user, WiFi, Telephony, Print, Cameras, etc) and the other two Ex4600 for Servers.
The Users' Core (It has configure BGP, VRRP, DHCP Relay, VLAN L2/L3, etc) and The Server's Core (It has configure BGP, VRRP, etc). All Cores are on going.

My question is: Can I configure MC-LAG on all ex4600?, MC-LAG is compatibility with all those feature? (BGP, VRRP, DHCP Relay, etc)

I understand that it will be affect the network, but not problem. 

Tks for your help!

MC-LAG is definitely not the same as a virtual chassis!

In a virtual chassis (just like a stack) one of the units has the control plane (Juniper calls it the "routing engine") and controls all of the hardware of all members of the virtual chassis, just like a physical chassis. This is very practical because you only have one unit to configure, there are never inconsistencies etc.

In an MC-LAG system, all of the members are running their own control plane. If you only have L3 connections this can be easily implemented using VRRP.

If you're running L2, It's common usage to connect access switches to two or more distribution switches at the same time (LAG/trunk). Without any precautions you will get massive loops if you simply connect the access switches to the all of the distribution switches. One counter measure is to use spanning tree and then always only one path is used. I consider this very outdated and waste of resources. Another approach is to use someting what Nortel/Avaya use to call split-multi-link-trunk (SMLT), Cisco calls it vPC and actually I don't know how Juniper calls it. It's MC-LAG. Even though all distribution (upstream) switches are connected and all run their own routing engine,  no loops are created because they share their (a.o) forwarding table. This is more prone to configuration errors but it is more resilient to bugs in firmware and is believed to yield a higher total uptime. So it's something  one needs to consider on a per-case situation.

Hi GRZEGORZ DACKA

I get it, ok tks!

Hello 

I don't really understand what MC-LAG has to do with bgp, dhcp etc.
You pin the EX4600 in virtual chassie and then you can use MC-LAG 

What and how you do on this lag is another matter 

Hi People,

Currently, I have four Cores separate for my internal network. Two ex4600 for Users (End-user, WiFi, Telephony, Print, Cameras, etc) and the other two Ex4600 for Servers.
The Users' Core (It has configure BGP, VRRP, DHCP Relay, VLAN L2/L3, etc) and The Server's Core (It has configure BGP, VRRP, etc). All Cores are on going.

My question is: Can I configure MC-LAG on all ex4600?, MC-LAG is compatibility with all those feature? (BGP, VRRP, DHCP Relay, etc)

I understand that it will be affect the network, but not problem. 

Tks for your help!

MC-LAG is definitely not the same as a virtual chassis!

In a virtual chassis (just like a stack) one of the units has the control plane (Juniper calls it the "routing engine") and controls all of the hardware of all members of the virtual chassis, just like a physical chassis. This is very practical because you only have one unit to configure, there are never inconsistencies etc.

In an MC-LAG system, all of the members are running their own control plane. If you only have L3 connections this can be easily implemented using VRRP.

If you're running L2, It's common usage to connect access switches to two or more distribution switches at the same time (LAG/trunk). Without any precautions you will get massive loops if you simply connect the access switches to the all of the distribution switches. One counter measure is to use spanning tree and then always only one path is used. I consider this very outdated and waste of resources. Another approach is to use someting what Nortel/Avaya use to call split-multi-link-trunk (SMLT), Cisco calls it vPC and actually I don't know how Juniper calls it. It's MC-LAG. Even though all distribution (upstream) switches are connected and all run their own routing engine,  no loops are created because they share their (a.o) forwarding table. This is more prone to configuration errors but it is more resilient to bugs in firmware and is believed to yield a higher total uptime. So it's something  one needs to consider on a per-case situation.

Hi GRZEGORZ DACKA

I get it, ok tks!

Hello 

I don't really understand what MC-LAG has to do with bgp, dhcp etc.
You pin the EX4600 in virtual chassie and then you can use MC-LAG 

What and how you do on this lag is another matter 

Hi People,

Currently, I have four Cores separate for my internal network. Two ex4600 for Users (End-user, WiFi, Telephony, Print, Cameras, etc) and the other two Ex4600 for Servers.
The Users' Core (It has configure BGP, VRRP, DHCP Relay, VLAN L2/L3, etc) and The Server's Core (It has configure BGP, VRRP, etc). All Cores are on going.

My question is: Can I configure MC-LAG on all ex4600?, MC-LAG is compatibility with all those feature? (BGP, VRRP, DHCP Relay, etc)

I understand that it will be affect the network, but not problem. 

Tks for your help!

Hi Erik Slagter,

Yes, Virtual Chassis is different that MC-LAG. I have experience with VC, but now I have opportunity for configure MC_LAG on my network ongoing.

Today the two ex4600 are configure as on the picture (on attachment), I have 03 sessions BGP, why I say if the MC-LAG is compatibility with BGP?
 I say it because when I'll configure the MC-LAG on ex4600, the iBGP is necessary? or not? 

Tks!

MC-LAG is definitely not the same as a virtual chassis!

In a virtual chassis (just like a stack) one of the units has the control plane (Juniper calls it the "routing engine") and controls all of the hardware of all members of the virtual chassis, just like a physical chassis. This is very practical because you only have one unit to configure, there are never inconsistencies etc.

In an MC-LAG system, all of the members are running their own control plane. If you only have L3 connections this can be easily implemented using VRRP.

If you're running L2, It's common usage to connect access switches to two or more distribution switches at the same time (LAG/trunk). Without any precautions you will get massive loops if you simply connect the access switches to the all of the distribution switches. One counter measure is to use spanning tree and then always only one path is used. I consider this very outdated and waste of resources. Another approach is to use someting what Nortel/Avaya use to call split-multi-link-trunk (SMLT), Cisco calls it vPC and actually I don't know how Juniper calls it. It's MC-LAG. Even though all distribution (upstream) switches are connected and all run their own routing engine,  no loops are created because they share their (a.o) forwarding table. This is more prone to configuration errors but it is more resilient to bugs in firmware and is believed to yield a higher total uptime. So it's something  one needs to consider on a per-case situation.

Hi GRZEGORZ DACKA

I get it, ok tks!

Hello 

I don't really understand what MC-LAG has to do with bgp, dhcp etc.
You pin the EX4600 in virtual chassie and then you can use MC-LAG 

What and how you do on this lag is another matter 

Hi People,

Currently, I have four Cores separate for my internal network. Two ex4600 for Users (End-user, WiFi, Telephony, Print, Cameras, etc) and the other two Ex4600 for Servers.
The Users' Core (It has configure BGP, VRRP, DHCP Relay, VLAN L2/L3, etc) and The Server's Core (It has configure BGP, VRRP, etc). All Cores are on going.

My question is: Can I configure MC-LAG on all ex4600?, MC-LAG is compatibility with all those feature? (BGP, VRRP, DHCP Relay, etc)

I understand that it will be affect the network, but not problem. 

Tks for your help!

Hello

I'm having trouble understanding what you're trying to accomplish
On ex4600 you want to run BGP with operator and download full feed ? This will not work you will not fit the array
The second thing is that it's hard to know what exactly you want to do because you give perfunctory information. 
I don't know what BGP has to do with MC-LAG compatibility.

One is L3 the other is L2

Hi Erik Slagter,

Yes, Virtual Chassis is different that MC-LAG. I have experience with VC, but now I have opportunity for configure MC_LAG on my network ongoing.

Today the two ex4600 are configure as on the picture (on attachment), I have 03 sessions BGP, why I say if the MC-LAG is compatibility with BGP?
 I say it because when I'll configure the MC-LAG on ex4600, the iBGP is necessary? or not? 

Tks!

MC-LAG is definitely not the same as a virtual chassis!

In a virtual chassis (just like a stack) one of the units has the control plane (Juniper calls it the "routing engine") and controls all of the hardware of all members of the virtual chassis, just like a physical chassis. This is very practical because you only have one unit to configure, there are never inconsistencies etc.

In an MC-LAG system, all of the members are running their own control plane. If you only have L3 connections this can be easily implemented using VRRP.

If you're running L2, It's common usage to connect access switches to two or more distribution switches at the same time (LAG/trunk). Without any precautions you will get massive loops if you simply connect the access switches to the all of the distribution switches. One counter measure is to use spanning tree and then always only one path is used. I consider this very outdated and waste of resources. Another approach is to use someting what Nortel/Avaya use to call split-multi-link-trunk (SMLT), Cisco calls it vPC and actually I don't know how Juniper calls it. It's MC-LAG. Even though all distribution (upstream) switches are connected and all run their own routing engine,  no loops are created because they share their (a.o) forwarding table. This is more prone to configuration errors but it is more resilient to bugs in firmware and is believed to yield a higher total uptime. So it's something  one needs to consider on a per-case situation.

Hi GRZEGORZ DACKA

I get it, ok tks!

Hello 

I don't really understand what MC-LAG has to do with bgp, dhcp etc.
You pin the EX4600 in virtual chassie and then you can use MC-LAG 

What and how you do on this lag is another matter 

Hi People,

Currently, I have four Cores separate for my internal network. Two ex4600 for Users (End-user, WiFi, Telephony, Print, Cameras, etc) and the other two Ex4600 for Servers.
The Users' Core (It has configure BGP, VRRP, DHCP Relay, VLAN L2/L3, etc) and The Server's Core (It has configure BGP, VRRP, etc). All Cores are on going.

My question is: Can I configure MC-LAG on all ex4600?, MC-LAG is compatibility with all those feature? (BGP, VRRP, DHCP Relay, etc)

I understand that it will be affect the network, but not problem. 

Tks for your help!

I don't agree. I am not too knowledgeable on BGP, but there is an important difference between mc-lag and vc here. A vc will have only one routing engine and therefore it will only have one ip address in all vlans *), no need to run vrrp or setting up a (BGP) relation between different members. On a mc-lag setup you will need to use vrrp and/or setup a bgp relationship and all members will act as a complete independent router (which happen to be in the same vlans).

Both approaches have pros and cons.

(this is also virtual, each member will forward traffic by itself if it can, it will not necessarily forward L2 or L3 traffic to the member running the routing engine, but you won't see that process).

Hello

I'm having trouble understanding what you're trying to accomplish
On ex4600 you want to run BGP with operator and download full feed ? This will not work you will not fit the array
The second thing is that it's hard to know what exactly you want to do because you give perfunctory information. 
I don't know what BGP has to do with MC-LAG compatibility.

One is L3 the other is L2

Hi Erik Slagter,

Yes, Virtual Chassis is different that MC-LAG. I have experience with VC, but now I have opportunity for configure MC_LAG on my network ongoing.

Today the two ex4600 are configure as on the picture (on attachment), I have 03 sessions BGP, why I say if the MC-LAG is compatibility with BGP?
 I say it because when I'll configure the MC-LAG on ex4600, the iBGP is necessary? or not? 

Tks!

MC-LAG is definitely not the same as a virtual chassis!

In a virtual chassis (just like a stack) one of the units has the control plane (Juniper calls it the "routing engine") and controls all of the hardware of all members of the virtual chassis, just like a physical chassis. This is very practical because you only have one unit to configure, there are never inconsistencies etc.

In an MC-LAG system, all of the members are running their own control plane. If you only have L3 connections this can be easily implemented using VRRP.

If you're running L2, It's common usage to connect access switches to two or more distribution switches at the same time (LAG/trunk). Without any precautions you will get massive loops if you simply connect the access switches to the all of the distribution switches. One counter measure is to use spanning tree and then always only one path is used. I consider this very outdated and waste of resources. Another approach is to use someting what Nortel/Avaya use to call split-multi-link-trunk (SMLT), Cisco calls it vPC and actually I don't know how Juniper calls it. It's MC-LAG. Even though all distribution (upstream) switches are connected and all run their own routing engine,  no loops are created because they share their (a.o) forwarding table. This is more prone to configuration errors but it is more resilient to bugs in firmware and is believed to yield a higher total uptime. So it's something  one needs to consider on a per-case situation.

Hi GRZEGORZ DACKA

I get it, ok tks!

Hello 

I don't really understand what MC-LAG has to do with bgp, dhcp etc.
You pin the EX4600 in virtual chassie and then you can use MC-LAG 

What and how you do on this lag is another matter 

Hi People,

Currently, I have four Cores separate for my internal network. Two ex4600 for Users (End-user, WiFi, Telephony, Print, Cameras, etc) and the other two Ex4600 for Servers.
The Users' Core (It has configure BGP, VRRP, DHCP Relay, VLAN L2/L3, etc) and The Server's Core (It has configure BGP, VRRP, etc). All Cores are on going.

My question is: Can I configure MC-LAG on all ex4600?, MC-LAG is compatibility with all those feature? (BGP, VRRP, DHCP Relay, etc)

I understand that it will be affect the network, but not problem. 

Tks for your help!

Hello

Maybe let's start the conversation with what the author wants to perform because that's how we have to hire a fortune teller.
On the one hand he wants MC-lag on the other is VC or not and then there is BGP 

I don't agree. I am not too knowledgeable on BGP, but there is an important difference between mc-lag and vc here. A vc will have only one routing engine and therefore it will only have one ip address in all vlans *), no need to run vrrp or setting up a (BGP) relation between different members. On a mc-lag setup you will need to use vrrp and/or setup a bgp relationship and all members will act as a complete independent router (which happen to be in the same vlans).

Both approaches have pros and cons.

(this is also virtual, each member will forward traffic by itself if it can, it will not necessarily forward L2 or L3 traffic to the member running the routing engine, but you won't see that process).

Hello

I'm having trouble understanding what you're trying to accomplish
On ex4600 you want to run BGP with operator and download full feed ? This will not work you will not fit the array
The second thing is that it's hard to know what exactly you want to do because you give perfunctory information. 
I don't know what BGP has to do with MC-LAG compatibility.

One is L3 the other is L2

Hi Erik Slagter,

Yes, Virtual Chassis is different that MC-LAG. I have experience with VC, but now I have opportunity for configure MC_LAG on my network ongoing.

Today the two ex4600 are configure as on the picture (on attachment), I have 03 sessions BGP, why I say if the MC-LAG is compatibility with BGP?
 I say it because when I'll configure the MC-LAG on ex4600, the iBGP is necessary? or not? 

Tks!

MC-LAG is definitely not the same as a virtual chassis!

In a virtual chassis (just like a stack) one of the units has the control plane (Juniper calls it the "routing engine") and controls all of the hardware of all members of the virtual chassis, just like a physical chassis. This is very practical because you only have one unit to configure, there are never inconsistencies etc.

In an MC-LAG system, all of the members are running their own control plane. If you only have L3 connections this can be easily implemented using VRRP.

If you're running L2, It's common usage to connect access switches to two or more distribution switches at the same time (LAG/trunk). Without any precautions you will get massive loops if you simply connect the access switches to the all of the distribution switches. One counter measure is to use spanning tree and then always only one path is used. I consider this very outdated and waste of resources. Another approach is to use someting what Nortel/Avaya use to call split-multi-link-trunk (SMLT), Cisco calls it vPC and actually I don't know how Juniper calls it. It's MC-LAG. Even though all distribution (upstream) switches are connected and all run their own routing engine,  no loops are created because they share their (a.o) forwarding table. This is more prone to configuration errors but it is more resilient to bugs in firmware and is believed to yield a higher total uptime. So it's something  one needs to consider on a per-case situation.

Hi GRZEGORZ DACKA

I get it, ok tks!

Hello 

I don't really understand what MC-LAG has to do with bgp, dhcp etc.
You pin the EX4600 in virtual chassie and then you can use MC-LAG 

What and how you do on this lag is another matter 

Hi People,

Currently, I have four Cores separate for my internal network. Two ex4600 for Users (End-user, WiFi, Telephony, Print, Cameras, etc) and the other two Ex4600 for Servers.
The Users' Core (It has configure BGP, VRRP, DHCP Relay, VLAN L2/L3, etc) and The Server's Core (It has configure BGP, VRRP, etc). All Cores are on going.

My question is: Can I configure MC-LAG on all ex4600?, MC-LAG is compatibility with all those feature? (BGP, VRRP, DHCP Relay, etc)

I understand that it will be affect the network, but not problem. 

Tks for your help!

Hi GRZEGORZ

Currently my network is configure as the picture on attachement (02 ex4600 with 03 session BGP (2xeBGP and 1xiBGP) . The 02 eBGP to ISP and iBGP between ex4600. So far so good?

When I'll configure MC-LAG  the iBGP is necessary? or not?

In my point of view yes It is necessary because MC-LAG is not Virtual Chassis.

That is my first question

Tks!

Hello

Maybe let's start the conversation with what the author wants to perform because that's how we have to hire a fortune teller.
On the one hand he wants MC-lag on the other is VC or not and then there is BGP 

I don't agree. I am not too knowledgeable on BGP, but there is an important difference between mc-lag and vc here. A vc will have only one routing engine and therefore it will only have one ip address in all vlans *), no need to run vrrp or setting up a (BGP) relation between different members. On a mc-lag setup you will need to use vrrp and/or setup a bgp relationship and all members will act as a complete independent router (which happen to be in the same vlans).

Both approaches have pros and cons.

(this is also virtual, each member will forward traffic by itself if it can, it will not necessarily forward L2 or L3 traffic to the member running the routing engine, but you won't see that process).

Hello

I'm having trouble understanding what you're trying to accomplish
On ex4600 you want to run BGP with operator and download full feed ? This will not work you will not fit the array
The second thing is that it's hard to know what exactly you want to do because you give perfunctory information. 
I don't know what BGP has to do with MC-LAG compatibility.

One is L3 the other is L2

Hi Erik Slagter,

Yes, Virtual Chassis is different that MC-LAG. I have experience with VC, but now I have opportunity for configure MC_LAG on my network ongoing.

Today the two ex4600 are configure as on the picture (on attachment), I have 03 sessions BGP, why I say if the MC-LAG is compatibility with BGP?
 I say it because when I'll configure the MC-LAG on ex4600, the iBGP is necessary? or not? 

Tks!

MC-LAG is definitely not the same as a virtual chassis!

In a virtual chassis (just like a stack) one of the units has the control plane (Juniper calls it the "routing engine") and controls all of the hardware of all members of the virtual chassis, just like a physical chassis. This is very practical because you only have one unit to configure, there are never inconsistencies etc.

In an MC-LAG system, all of the members are running their own control plane. If you only have L3 connections this can be easily implemented using VRRP.

If you're running L2, It's common usage to connect access switches to two or more distribution switches at the same time (LAG/trunk). Without any precautions you will get massive loops if you simply connect the access switches to the all of the distribution switches. One counter measure is to use spanning tree and then always only one path is used. I consider this very outdated and waste of resources. Another approach is to use someting what Nortel/Avaya use to call split-multi-link-trunk (SMLT), Cisco calls it vPC and actually I don't know how Juniper calls it. It's MC-LAG. Even though all distribution (upstream) switches are connected and all run their own routing engine,  no loops are created because they share their (a.o) forwarding table. This is more prone to configuration errors but it is more resilient to bugs in firmware and is believed to yield a higher total uptime. So it's something  one needs to consider on a per-case situation.

Hi GRZEGORZ DACKA

I get it, ok tks!

Hello 

I don't really understand what MC-LAG has to do with bgp, dhcp etc.
You pin the EX4600 in virtual chassie and then you can use MC-LAG 

What and how you do on this lag is another matter 

Hi People,

Currently, I have four Cores separate for my internal network. Two ex4600 for Users (End-user, WiFi, Telephony, Print, Cameras, etc) and the other two Ex4600 for Servers.
The Users' Core (It has configure BGP, VRRP, DHCP Relay, VLAN L2/L3, etc) and The Server's Core (It has configure BGP, VRRP, etc). All Cores are on going.

My question is: Can I configure MC-LAG on all ex4600?, MC-LAG is compatibility with all those feature? (BGP, VRRP, DHCP Relay, etc)

I understand that it will be affect the network, but not problem. 

Tks for your help!

Hi,

Yes, IBGP is needed. I think u need simulate first using vMX so u will know how the traffic flow failover.


Thanks

Hi GRZEGORZ

Currently my network is configure as the picture on attachement (02 ex4600 with 03 session BGP (2xeBGP and 1xiBGP) . The 02 eBGP to ISP and iBGP between ex4600. So far so good?

When I'll configure MC-LAG  the iBGP is necessary? or not?

In my point of view yes It is necessary because MC-LAG is not Virtual Chassis.

That is my first question

Tks!

Hello

Maybe let's start the conversation with what the author wants to perform because that's how we have to hire a fortune teller.
On the one hand he wants MC-lag on the other is VC or not and then there is BGP 

I don't agree. I am not too knowledgeable on BGP, but there is an important difference between mc-lag and vc here. A vc will have only one routing engine and therefore it will only have one ip address in all vlans *), no need to run vrrp or setting up a (BGP) relation between different members. On a mc-lag setup you will need to use vrrp and/or setup a bgp relationship and all members will act as a complete independent router (which happen to be in the same vlans).

Both approaches have pros and cons.

(this is also virtual, each member will forward traffic by itself if it can, it will not necessarily forward L2 or L3 traffic to the member running the routing engine, but you won't see that process).

Hello

I'm having trouble understanding what you're trying to accomplish
On ex4600 you want to run BGP with operator and download full feed ? This will not work you will not fit the array
The second thing is that it's hard to know what exactly you want to do because you give perfunctory information. 
I don't know what BGP has to do with MC-LAG compatibility.

One is L3 the other is L2

Hi Erik Slagter,

Yes, Virtual Chassis is different that MC-LAG. I have experience with VC, but now I have opportunity for configure MC_LAG on my network ongoing.

Today the two ex4600 are configure as on the picture (on attachment), I have 03 sessions BGP, why I say if the MC-LAG is compatibility with BGP?
 I say it because when I'll configure the MC-LAG on ex4600, the iBGP is necessary? or not? 

Tks!

MC-LAG is definitely not the same as a virtual chassis!

In a virtual chassis (just like a stack) one of the units has the control plane (Juniper calls it the "routing engine") and controls all of the hardware of all members of the virtual chassis, just like a physical chassis. This is very practical because you only have one unit to configure, there are never inconsistencies etc.

In an MC-LAG system, all of the members are running their own control plane. If you only have L3 connections this can be easily implemented using VRRP.

If you're running L2, It's common usage to connect access switches to two or more distribution switches at the same time (LAG/trunk). Without any precautions you will get massive loops if you simply connect the access switches to the all of the distribution switches. One counter measure is to use spanning tree and then always only one path is used. I consider this very outdated and waste of resources. Another approach is to use someting what Nortel/Avaya use to call split-multi-link-trunk (SMLT), Cisco calls it vPC and actually I don't know how Juniper calls it. It's MC-LAG. Even though all distribution (upstream) switches are connected and all run their own routing engine,  no loops are created because they share their (a.o) forwarding table. This is more prone to configuration errors but it is more resilient to bugs in firmware and is believed to yield a higher total uptime. So it's something  one needs to consider on a per-case situation.

Hi GRZEGORZ DACKA

I get it, ok tks!

Hello 

I don't really understand what MC-LAG has to do with bgp, dhcp etc.
You pin the EX4600 in virtual chassie and then you can use MC-LAG 

What and how you do on this lag is another matter 

Hi People,

Currently, I have four Cores separate for my internal network. Two ex4600 for Users (End-user, WiFi, Telephony, Print, Cameras, etc) and the other two Ex4600 for Servers.
The Users' Core (It has configure BGP, VRRP, DHCP Relay, VLAN L2/L3, etc) and The Server's Core (It has configure BGP, VRRP, etc). All Cores are on going.

My question is: Can I configure MC-LAG on all ex4600?, MC-LAG is compatibility with all those feature? (BGP, VRRP, DHCP Relay, etc)

I understand that it will be affect the network, but not problem. 

Tks for your help!

Hello

Maybe let's start the conversation with what the author wants to perform because that's how we have to hire a fortune teller.
On the one hand he wants MC-lag on the other is VC or not and then there is BGP 

I don't agree. I am not too knowledgeable on BGP, but there is an important difference between mc-lag and vc here. A vc will have only one routing engine and therefore it will only have one ip address in all vlans *), no need to run vrrp or setting up a (BGP) relation between different members. On a mc-lag setup you will need to use vrrp and/or setup a bgp relationship and all members will act as a complete independent router (which happen to be in the same vlans).

Both approaches have pros and cons.

(this is also virtual, each member will forward traffic by itself if it can, it will not necessarily forward L2 or L3 traffic to the member running the routing engine, but you won't see that process).

Hello

I'm having trouble understanding what you're trying to accomplish
On ex4600 you want to run BGP with operator and download full feed ? This will not work you will not fit the array
The second thing is that it's hard to know what exactly you want to do because you give perfunctory information. 
I don't know what BGP has to do with MC-LAG compatibility.

One is L3 the other is L2

Hi Erik Slagter,

Yes, Virtual Chassis is different that MC-LAG. I have experience with VC, but now I have opportunity for configure MC_LAG on my network ongoing.

Today the two ex4600 are configure as on the picture (on attachment), I have 03 sessions BGP, why I say if the MC-LAG is compatibility with BGP?
 I say it because when I'll configure the MC-LAG on ex4600, the iBGP is necessary? or not? 

Tks!

MC-LAG is definitely not the same as a virtual chassis!

In a virtual chassis (just like a stack) one of the units has the control plane (Juniper calls it the "routing engine") and controls all of the hardware of all members of the virtual chassis, just like a physical chassis. This is very practical because you only have one unit to configure, there are never inconsistencies etc.

In an MC-LAG system, all of the members are running their own control plane. If you only have L3 connections this can be easily implemented using VRRP.

If you're running L2, It's common usage to connect access switches to two or more distribution switches at the same time (LAG/trunk). Without any precautions you will get massive loops if you simply connect the access switches to the all of the distribution switches. One counter measure is to use spanning tree and then always only one path is used. I consider this very outdated and waste of resources. Another approach is to use someting what Nortel/Avaya use to call split-multi-link-trunk (SMLT), Cisco calls it vPC and actually I don't know how Juniper calls it. It's MC-LAG. Even though all distribution (upstream) switches are connected and all run their own routing engine,  no loops are created because they share their (a.o) forwarding table. This is more prone to configuration errors but it is more resilient to bugs in firmware and is believed to yield a higher total uptime. So it's something  one needs to consider on a per-case situation.

Hi GRZEGORZ DACKA

I get it, ok tks!

Hello 

I don't really understand what MC-LAG has to do with bgp, dhcp etc.
You pin the EX4600 in virtual chassie and then you can use MC-LAG 

What and how you do on this lag is another matter 

Hi People,

Currently, I have four Cores separate for my internal network. Two ex4600 for Users (End-user, WiFi, Telephony, Print, Cameras, etc) and the other two Ex4600 for Servers.
The Users' Core (It has configure BGP, VRRP, DHCP Relay, VLAN L2/L3, etc) and The Server's Core (It has configure BGP, VRRP, etc). All Cores are on going.

My question is: Can I configure MC-LAG on all ex4600?, MC-LAG is compatibility with all those feature? (BGP, VRRP, DHCP Relay, etc)

I understand that it will be affect the network, but not problem. 

Tks for your help!