Hello Alan
Your thinking is correct. vPC is the Nexus methodology of creating redundancy between two Layer 3 switches. And yes, you can use HSRP/VRRP to employ layer 3 gateway redundancy for your access layer VLANs.
Now your question concerning how many access switches can you have in such a topology is a very good one. The only strict limitation here is the number of ports your Nexus switches have. If you have 24 ports on each switch for example, and four ports are being used for uplinks and your vPC connections, then the other 20 ports can all be used to connect to access switches. However, is that acceptable? Well it depends.
We must look here at the concept of oversubscription ratio. This is the ratio between the total downlink bandwidth and the total uplink bandwidth. For example, a distribution switch may have 48 1Gbps connected to access switches (downlinks), and one 10 Gbps port connected to the core layer (uplinks). The total aggregate bandwidth of the downlinks is 48 Gbps while the total aggregate uplink is 10 Gbps. The oversubscription ratio is 48:10 or 4.8:1.
This means that you are provisioning total downlink bandwidth at 4.8 times the total uplink bandwidth. This means that if all downlink connections were being used at full capacity at the same time, the uplink bandwidth would not be able to handle this traffic, and this is why the word “oversubscription” is used. However, this design depends upon the fact that this is a very rare occurrence, and in general, downlinks will never use full bandwidth all at the same time.
A general rule of thumb is that data oversubscription for access ports on the access to distribution uplink should not exceed 20:1 while from the distribution to the core links, this should not exceed 4:1.
So to answer your question, you can connect as many access switches as you like as long as your total aggregate uplink bandwidth to total aggregate downlink bandwidth is a ratio of 20:1 or less.
I hope this has been helpful!
Laz