# Rearange input data - unknown size of input

I have some data of form showed below to squezze through some neural network.

Customer Step Var0  ... VarN
0        0    some  ... stuff
0        1    again ... stuff
0        2    foo   ... bar
0        3    bla   ... blub
1        0    other ... stuff
1        1    and   ... ongoing
2        0    and   ... so on
.        .    .     ... .
.        .    .     ... .
.        .    .     ... .


As the neural network has a fixed input layer size, but the customers step count can differ at each customer, I have following question:

What is the best practise (if exists) to rearange the given data for using as input of a NN?