Hmm, I never have dictated to anyone whether on the customer side as an architect, manager, director or now on the pre-sales engineering size.   I can inform and advise based upon my experience and knowledge gathered from my fellow pre and post sales engineers.   I agree that the cost of a solution has to meet the budgetary, security, and operational availability requirements of the project / customer.  The engineers that work for the organization and for vendors / VARs need to take those above items into account when creating a design.  A network infrastructure cannot exceed the budget, or be a massive overkill for what the business requires.   A Catalyst 6807 with quad core, Nexus 7700/9500 with dual supervisors, etc. is, of course, for large enterprise campus or data center cores that support significant numbers of users, storage, compute, or remote sites, etc. is a solution that should be considered when the highest level of uptime and stability are a requirement. ... View more

With the Catalyst 9400 not supporting VSS, how could it be considered ready for a core?  The roadmap for it having VSS is a year or so, correct?     Also, with the Cat9Ks only shipping for a few months, and very limited numbers of deployments, and being deployed in a production environment for a very short time frame, I would be tentative about recommending the C9Ks for a production environment where high up time is a requirement.   There are limited releases of code (16.6.1 & 16.6.2) available for the C9Ks, which is a limiting factor to consider if a bug is encountered.  With more seasoned platforms, there are long lived code trains with recommended releases.   Stackwise virtual is new, again with limited deployment and not supported on all models.  Just the 9500-24Q at the moment.   The Cisco CVDs and design guides make no reference to Catalyst 9Ks, they recommend the Catalyst 6807 or Nexus 9K.     With Quad Supervisors, on a pair of Cat6Ks or N7Ks, software upgrades or supervisor failures can occur with virtually no downtime, perhaps a few pings will drop.   (A properly designed and configured implementation is required, and there are certain upgrades that could be more challenging-if the N7k lines cards need an EPLD upgrade for example).   However, Cat6Ks or N7Ks have been rock solid for years, have all of the high availability features that an enterprise network requires, and have been proven to deliver superior up time, and resilience during component failures and upgrades.   Upgrading a Cat9K, or a supervisor failure is going to be a much more significant outage without VSS, ISSU, and the limitation from the 16.6.2 release notes I reference below-several minutes I believe.    **************** "Redundancy—The supervisor module (hardware) supports redundancy. Software redundancy is supported in IOS XE Everest 16.6.2. The associated route processor redundancy (RPR) feature is currently not supported."   https://www.cisco.com/c/en/us/td/docs/switches/lan/catalyst9400/software/release/16-6/release_notes/ol-16-6-9400.html ***************   Regards, Jason ... View more