Is it possible to replicaset and shard a collection?

I've seen that for making a data distribution i can choose between ReplicaSet or Sharding. Using a mongodb code with those parameters --shardsvr --replSet.

mongod --shardsvr --port 27021 --replSet rs1 --dbpath C:\data\data5 --bind_ip localhost

What i'm doing is making a shard server that is also a replicaSet or i'm just doing a shard? And if i'm not doing both what should i write?

Edit

mongod --shardsvr --port 27020 --replSet rs1 --dbpath C:\data\data4 --bind_ip localhost
mongod --shardsvr --port 27021 --replSet rs1 --dbpath C:\data\data5 --bind_ip localhost
mongod --shardsvr --port 27022 --replSet rs1 --dbpath C:\data\data6 --bind_ip localhost
mongos --port 40000 --configdb rs0/localhost:27017
mongo --port 40000
rs.initiate( { _id : "rs1",  members: [   { _id: 0, host: "localhost:27020" },   { _id: 1, host: "localhost:27021" },   { _id: 2, host: "localhost:27022" }    ] })
sh.addShard( "rs1/localhost:27020,localhost:27021,localhost:27022")
sh.enableSharding("db")
sh.shardCollection("db.collection", { "_id" : 1 } )

When i do this kind of a code what i get is a collection sharded by 3 "servers" that are sharding and repliceSet aswell? And with mongos on 40'000 port i can interface and make querys?

Solution

Replica sets and sharding serve different purposes:

replica sets provide redundancy (if one node fails, data is also on other nodes)
sharding provides scalability (if you have too much data for one server, you can spread it over multiple servers)

When you set up a sharded cluster, each shard can be (in theory) backed by a single node or a replica set. MongoDB currently requires that shards are backed by replica sets, in older versions single nodes were also permitted.

Therefore:

Sharded cluster with single nodes = no redundancy. Each document is stored once, if any single node fails you lose data. But you can get scalability based on number of shards you deploy.
Sharded cluster with backing replica sets = both scalability and redundancy.