We are using cassandra-1.1.
For some optimization purpose we decide to increase no of column families in our keyspace.
Will it have any impact on heap utilizatio
Recent versions of Cassandra allocate a minimum of 1MB on the heap for each column family, so you can treat that as the lower bound for heap consumption. Bloom filters also take up heap space in a way that doesn't necessarily depend on how much you use the column family.
Are you talking about going from 5 to 10 column families? Or 10 to 1000? You certainly can run out of heap space with 10 or 1000 column families, it just depends a lot on the rate at which you're inserting data.