Dual rate, 3 color, can be used with a policer or shaper. Policing does not buffer any traffic.
Seldom is there a need to police ingress traffic, unless you're enforcing a bandwidth limit, as your ISP might be doing to you.
Unlikely there's any good reason for you to police ingress traffic from the Internet. There are some extreme cases, excluding contracted bandwidth, but those often require expert QoS knowledge to understand and configure (and their effectiveness is often not nearly as good as managing the other side's egress).
Again, shaping egress has some useful reasons for doing so, but also again, do you know them? Reason I ask, if ISP only allows 30 Mbps "up" and you shape for 30 Mbps "up" what do you expect to accomplish? 30 Mbps is still 30 Mbps, right? (It is and it isn't, it's somewhat an advanced QoS subject.)
Anyway, I'll describe 3 good reasons to use a shaper in a case like yours.
1), it can make it easier for you to know you're exceeding your "up" bandwidth, including by how much and how often. Otherwise, you submit traffic to ISP at 100 Mbps (correct?) and it's somewhat "invisible", to you, your ISP's treatment of overrate traffic. (Keep in mind, ideally, with a shaper, your ISP will no longer see any overrate traffic.)
2), assuming ISP is policing the 30 Mbps, a shaper provides a chance that traffic that would be dropped by a policer will not be dropped. If you're wondering, how so, whether using a policer or shaper, on a 100 Mbps interface, traffic is physically always sent at 100 Mbps, so how do we get a lessor "upload" speed?
Both a policer or shaper "count" volume of traffic transmitted during some time period. So for 30 Mbps, if the transmission of 100 Mbps is for more than 30% of the measured time period, the overrate/excess traffic is subject to policer/shaper enforcement. For a policer, overate traffic is dropped. For a shaper, overrate traffic is queued. Since a shaper queues excess traffic, if the shaping queue doesn't overflow, it's possible, later time periods that don't hit their maximum usage, can dequeue traffic, up to their time period's maximum.
3), since a shaper queues excess traffic, we can "manage" shaper queued traffic. For example, we can select dequeuing and/or dropping priorities.
Of the 3 reasons, above, their benefits generally increase sequentially, too, but how beneficial they are is much dependent on the actual case. For example, if your upload traffic never exceeds 30% measured time period utilization, none of the above would offer any benefit. Conversely, if you want to do VoIP and/or multiplayer on-line games concurrently with a cloud backup of your PC, you likely very, very much benefit from the #3, above.
