Drops discards are caused by (only caused by) not enough egress bandwidth.

Either the link is constrained by the port speed on the other side, or you have something enabled that activates policing (artificial bandwidth constraints).

As an example, suppose you have a 24 port switch, and on 20 of the ports you have PCs all operating a full duplex 100 Mbps link to the switch. On one of other open ports, you connect a file server/database server and it has a link operating at 10Mbps. All of the PCs are actively querying the server.

At some point, there is no place to send the traffic backed up waiting for the server's link to clear enough to send the data (all the switch's buffers are full), so it begins dropping, or in some cases, not accepting, new frames.

In the case of policing, you have enabled, either explicitly or as a default value, some threshold that says any traffic (possibly with a qualifier "of this type") that exceeds this level should be dropped.

Drops and discards are a congestion handling mechanism.

BTW: this server example also applies to an Internet / WAN connection ... having a bunch of machines at 100Mbps or Gig all trying to access a 1.5, 3, 6, 10 Mbps WAN link likely to show the same kind of activity.

Good Luck

Scott