IPv4 – Why Is the Don’t Fragment Flag Needed?

To me, it doesn't seem like it was necessarily a primary design choice to make the fragmentation offset on 8 byte boundaries. Rather, it was a result of the number of bits available to them within the primary design choices of how large the header should be, and how many bits were needed for other fields.

The designers chose the header size to be 20 bytes, and in that 20 bytes, given how they allocated bits for the various fields, they ended up with only 13 bits available for fragment offset.

So that gives 2^13 = 8192 possible offset values. At the same time, the total length field is 16 bits, so technically the max payload size is 2^16 - 20 = 65516 bytes.

Of course the fragment offset field must be able to span the entire spectrum of possible sizes, so 65536 / 8192 = 8.

This is, as Ricky Beam indicates, almost certainly TCP segmentation being done by the NIC. If you're capturing traffic on the server, capturing packets being sent by the server, and TCP segmentation offloading is being done, the server's operating system will hand packets bigger than the network's MTU to the NIC (with the expectation that the NIC will turn them into multiple IP packets) and to the packet capture mechanism and thus to the packet capture program, so it will appear as if a packet bigger than the network's MTU is being transmitted on that network - for example, Ethernet packets larger than 1514 bytes.