Just reading
http://miercom.com/pdf/reports/20150313.pdf
I think their comparison is poorly done: they have tested the throughput on a 1-U Cisco 4500 card against a 1/2-U HP card.
The comparison *should* be:
1x Cisco card (1U in the chassis) = 12 ports, 4 groups of 3 ports - each group max 12Gb, max 48Gb for the U, average max 4Gb per 10Gb port
2x HP card = 16 ports, 4 groups of 4 ports - each group max 23Gb, max 94Gb for the U, average max 6Gb per 10Gb port.
Essentially, the HP is 50% superior in terms of over-subscription.
Then they also get port density wrong, by not mentioning the rackspace each uses:
the Cisco supports 80 10Gb ports using 14U, 40 of them at Line Rate
the HP supports 96 10Gb ports using 7U, 48 of them at Line Rate
Then, they mention the 4510 can use VSS to create a 2-chassis pair.
They FAIL to mention you can use VSF to do the same thing with the 5412.
The HP delivers 20% MORE ports, 20% MORE at Line Rate, in ONE HALF the rackspace.
In terms of density and throughput, this document severely downplays the large advantages that the HP switch has over the 4510.
The issue of the much smaller buffers on the HP switch is a big difference. The 5412 is not a datacentre switch. As an Access switch, it is far superior to the Cisco in every relevant respect, and we haven't even spoken of cost. Off the top of my head, I would say the 5412 is half the price of the 4510.