The most likely cause for the lower performance after migrating to synchronously replicated storage is an increase in latency. Synchronous replication requires that writes are confirmed on both the primary and secondary storage systems before being acknowledged, which can add significant latency to the data processing. This added latency can slow down application performance even if CPU utilization and available memory do not indicate any immediate bottlenecks.
The most suitable method to meet the requirements of allowing HTTPS/HTTP traffic and disabling FTP and SMB traffic between the finance and marketing departments is to configure an Access Control List (ACL) between the VLANs. ACLs can be used to permit or deny traffic based on protocol and ports, ensuring that only the desired protocols (HTTPS and HTTP) are allowed while blocking any unwanted traffic (FTP and SMB).
The scenario described involves a cluster of virtual hosts that share storage internally, without an external storage server. Hyperconverged storage integrates compute, storage, and networking into a single system and allows for the creation of a virtual pool of storage that can be shared among the hosts. This matches the description provided in the question.

The current private cloud setup has 30 vCPUs and 512GB of RAM available. For the expected maximum sessions of close to 24,000, the application requires 8 nodes with 8 vCPUs and 128GB of RAM each. This amounts to a total requirement of 64 vCPUs and 1024GB of RAM, which the private cloud cannot support on its own. However, the average sessions are approximately 5,000, which requires 4 nodes with 6 vCPUs and 64GB of RAM each, amounting to 24 vCPUs and 256GB of RAM, which is within the current capacity. Setting up cloud bursting allows the private cloud to handle the average load while leveraging additional public cloud resources to manage peak loads, thus providing a quick and financially beneficial solution.
The alert regarding the license being out of compliance after upgrading the processors suggests that the licensing is based on the number of cores in the system. Core-based licensing typically requires a license for each core in the processor. When additional cores are added through an upgrade, it can cause the system to go out of compliance if there aren't enough licenses to cover the new core count.