If you are familiar with Microsoft NLB, you will know how much of a headache it can be for network engineers if it isn't configured properly.
Switches use MAC addresses to learn which devices are connected to which port, if a switch isn't able to correlate this information, it will flood traffic down all ports. I'm sure you can see the problem here, your expensive switch, is effectively a hub.
Why does it happen? Microsoft NLB uses a multicast MAC address, and switches aren't able learn the MAC address of the NLB IP, because the NLB never initiates traffic from the multicast MAC address, response traffic comes direct from the server.
Cisco support features which allows you to nail down a MAC address, to a specific port. There is no direct feature in Aruba / HPE Switches, but there is an alternative, and it does work.
ip arp-mcast-replies
Unfortunately, this isn't the full story. You also have to ensure that IGMP is enabled on the switch, and that multicast is configured as the 'cluster operation mode' in Microsoft NLB,
vlan 10
name "Client VLAN"
ip igmp
exit
Once you've configured these two aspects, you will no longer receive multicast traffic destined to an NLB address on all ports. I recommend you configure it on all switches on your campus.
Cisco Commands to achieve this:
arp x.x.x.x 0000.0000.0000 arpa
mac-address-table static Gix/x/x 0000.0000.0000