Lightboard Lessons: SSL Transactions Per Second
@bmichels2, great question! The answer depends on what version you are running. Things change as soon as you go into 12.x because you get per-guest SSL rate-limiting capabilities. Here are the two answers based on that split:
Pre 12.x: The guests are all in a round-robin method of sharing SSL resources. If the 8 vCPU guest is the only one accessing the SSL offload system, it can take all of the TPS the blade can provide. Otherwise, it’ll be sharing with as many other guests that are simultaneously accessing the SSL system, regardless of guest size.
12.x and later: There are 3 modes that you can have a guest in: Shared, Dedicated, and None. None, is just that, no hardware acceleration. Shared is the same as the round-robin method described above. Dedicated is where it gets interesting and, possibly, the situation you are in. When you have a guest in Dedicated SSL mode, the number of TPS is proportional to the number of vCPUs allocated to the guest. Since the guest in question is on a B2250, there are 20 vCPUs possible (1 for each HT, so a physical core is 2 vCPUs). Therefore, the amount of SSL TPS that this guest can have is 17.6K TPS, since the blade is rated for 44K SSL TPS (RSA) and 8 vCPUs is 40% of the total of 20 vCPUs: 2200 TPS/vCPU * 8 vCPUs == 17600 TPS.
I hope this helps!