SrikantRajeev
asked on
NSX Design
Attached is the current NSX network design.
I have L2 Switch connecting to the L3 Core switch. The ESXi hosts with NSX are connecting to the L2 switch.
OSPF is running between the L3 Core switch & the NSX.
In the near future I will be running short of the L2 switch ports where the ESXi hosts are connected.
For this I have 2 options
Option 1: Setup 2 more L2 switches parallel to the existing & connect the same to the L3 core switches. Similar to the existing setup.
Option 2: Setup 2 more L2 switches below the existing L2 switches & connect the ESXi hosts to the new L2 switches.
Would like to know which option is the best considering the pros & cons.
NSX_Design.png
I have L2 Switch connecting to the L3 Core switch. The ESXi hosts with NSX are connecting to the L2 switch.
OSPF is running between the L3 Core switch & the NSX.
In the near future I will be running short of the L2 switch ports where the ESXi hosts are connected.
For this I have 2 options
Option 1: Setup 2 more L2 switches parallel to the existing & connect the same to the L3 core switches. Similar to the existing setup.
Option 2: Setup 2 more L2 switches below the existing L2 switches & connect the ESXi hosts to the new L2 switches.
Would like to know which option is the best considering the pros & cons.
NSX_Design.png
The issue with option 2 is that as you add more ESXi hosts to the new switches you are limited to two uplinks to the layer 3 core. So depending on the utilization and how you have the network setup this could cause over subscription of the uplinks. So in general option 2 does not scale.
Depending on the utilization of the uplinks and the class of layer 2 switch that you are using, you may be able to use option 2, but why not connect right to the L3 core with new switches? So the answer is it depends on what the ESXi hosts are doing, how much capacity is available, etc .....
harbor235 ;}
Depending on the utilization of the uplinks and the class of layer 2 switch that you are using, you may be able to use option 2, but why not connect right to the L3 core with new switches? So the answer is it depends on what the ESXi hosts are doing, how much capacity is available, etc .....
harbor235 ;}
Without knowing more about the setup, I would say that option #1 should allow you more growth with out affecting performance. It depends on how much traffic from the ESX hosts actually has to go to the core.
If go with option#2 and you have a lot of traffic that goes from the ESX servers to the core, you will be increasing the amount of traffic that is flowing on the connections between your two existing L2 switched and your core. As harbor235 stated, this may not affect you at all depending on current utilization of those links.
The only downside I can think of with option #1 is that you could need more ports on your cores.
If go with option#2 and you have a lot of traffic that goes from the ESX servers to the core, you will be increasing the amount of traffic that is flowing on the connections between your two existing L2 switched and your core. As harbor235 stated, this may not affect you at all depending on current utilization of those links.
The only downside I can think of with option #1 is that you could need more ports on your cores.
OSPF is not relevant in a single L3 IP subnet
ASKER
Is there any document or design guide from vmware regarding how to scale the NSX design.
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
Thanks
In order to understand what may be best we (at least I do) would need to understand your L2 and L3 setup better. As an example. If you really only have a single subnet (x.x.x.0/22) that would imply a single L2 network. If you have a single L2 and L3 network, why are you running OSPF?