Single or multiple Distributed Switch deployment - use cases for both (pro's and con's)
Good day everyone,
As per the thread heading, I'm trying to cover all angles of what the pro's and con's are for a single and or multi distributed switch deployment. Currently we have a multi vDS deployment (one switch per cluster) which is used to address the 'not all eggs in one basket' should something go wrong, but also from an upgrade perspective that it can be tackled in a staged approach (upgrading VC, host, switch etc).
I've discovered that the aforementioned approach makes life for automation very difficult as there are certain 'limitations' in products such as vRA 6.x if you use an element called 'virtualmachine.network0.name'. For those who know, this is the first port group assignment of the first NIC of a VM and you cannot use custom names to have variations thereof (unless you have fancy coding skills in orchestrator) i.e 'custom.virtualmachine.network0.name'. All of this can be overcome if you have fancy coding skills in Orchestrator and port groups are dynamically populated via selection for dynamic placement.
I'd like to know from yourselves what you have found to be the pro's and con's of the above mentioned and what other scenarios could / could not justify a single and or multi vDS deployment.,
Also, what sort of code could I potentially look at (for Orchestrator) to enumerate port groups? I've been playing around with the XML xpression builder for use in vRA 6.x but again it leans more towards the use of a single vDS to eliminate extreme complexity.