NSX and SRM Configuration

[biosb]

Hello dear,

IN NSX with SRM Configuration:

1- Do we need to change VMkernel configuration (assigning it to Logical SW in VDS for each host) when we use NSX Cross vCenter with SRM?

2- Do you have step by step configuration document for NSX Cross vCenter with SRM?

And  I have another question,

IN NSX:

3- Can we just separate the VMs traffics using the separate Logical Switch? If not, How to separate and send traffics in Logical Switch to different "uplinks" that you want? (for example replication traffic)

[Sreec · Accepted solution]

IN NSX with SRM Configuration:

1- Do we need to change VMkernel configuration (assigning it to Logical SW in VDS for each host) when we use NSX Cross vCenter with SRM?

I'm unsure what you are referring ? Logical switches are for workloads.

2- Do you have step by step configuration document for NSX Cross vCenter with SRM?

You can refer NSX-V Multi-site Options and Cross-VC NSX Design Guide  . It's not a step by step guide. Document is specific to design decision . Form SRM perspective all we do is , align the right logical switch in the protection plan.

And  I have another question,

IN NSX:

3- Can we just separate the VMs traffics using the separate Logical Switch? If not, How to separate and send traffics in Logical Switch to different "uplinks" that you want? (for example replication traffic)

What are you trying to achieve ?  Different logical switches certainly provide an isolation, however that might not be enough based on the use case if you are looking from End-End design. Replication traffic should be kept aside from overlay network , unless you have a good bandwidth , don't go for such design.

[biosb]

hello dear Sreec

thank you for your response,

In your answer (my question 3), "Replication traffic should be kept aside from overlay network." please give me a reference or link for this.

And

I ask my Previous questions in a different way:

I have separated my traffics and transfer it to separate Uplinks (Uplink1 has connected to physical SW1 and Uplink 2 has connected to physical SW2). SW1 is for workloads and SW2 is for Replication traffic.(SW1 and SW2 don't have any connectivity). Now if I use "Logical Switch" for Replication VMkernel,

Can I separate Workload from Replication traffic and transfer that traffic to the Uplink that I want? or all of the traffics pass through the physical SW1 (that is for workload traffic)?

Thank you;

[Sreec]

In your answer (my question 3), "Replication traffic should be kept aside from overlay network." please give me a reference or link for this.

And

I'm unsure if there are any reference links for different types of traffic and how it should be segregated. My point is specific to how it will help to design an overlay network which doesn't get congested with other traffic .You can certainly club as many as traffic types via multiple uplinks , by default there is no guarantee which traffic will get more bandwidth. For the same reason we might end up with fine tuning with supported features . Feel free to read Overlay Design for NSX-T Data Center for the Management Domain

I ask my Previous questions in a different way:

I have separated my traffics and transfer it to separate Uplinks (Uplink1 has connected to physical SW1 and Uplink 2 has connected to physical SW2). SW1 is for workloads and SW2 is for Replication traffic.(SW1 and SW2 don't have any connectivity). Now if I use "Logical Switch" for Replication VMkernel,

Can I separate Workload from Replication traffic and transfer that traffic to the Uplink that I want? or all of the traffics pass through the physical SW1 (that is for workload traffic)?

Well, this approach gives you the flexibility of sending traffic via specific uplink to switch. You loose availability for both the traffic types. uplinks&switches are single point of failure.