Simple algorithm that returns the same shard irrespective of the number of new boxes added.

same_shard.py

'''
Question:
    Assume you want to shard big SQL user table. How will you obtain the same shard even after increasing the sql boxes.
    
Answer:
    Same shard number can be obtained if the current system state is preserved before adding new boxes.
    
NOTE: Code below requires python 2.6 or 2.7
'''

MAX_USER_ID = 0# for demonstration. This value is normally obtained from DB or some global state.
total_boxes = 5 #initial number of boxes
system_states = [] # maintains system state as tuple (range,boxes).

def add_boxes(n):
    global total_boxes
    system_states.append((MAX_USER_ID,total_boxes)) # preserving system state before adding boxes
    total_boxes += n

def get_shard(userId):# userid is assumed to be a sequentially increasing integer value.
    for state in system_states:
        userRange,boxes = state
        if userId <= userRange: # check previous system states and use that to compute shard
            return userId % boxes
    return userId % total_boxes # this is a new userid whose value has not been seen yet. Use the current set of boxes to compute shard

if __name__ == '__main__':

    print get_shard(999)
    print get_shard(1000)
    print get_shard(1001)

    MAX_USER_ID = 1001 #we have added 1001 users so far.
    add_boxes(3)#adding 3 more boxes
    print 'After adding 3 more boxes'
    print get_shard(999)
    print get_shard(1000)
    print get_shard(1001)
    print get_shard(2000)
    print get_shard(2007)

    MAX_USER_ID = 2007 #we have added 2007 users so far.
    add_boxes(5)
    print 'After adding 5 more boxes'
    print get_shard(999)
    print get_shard(1000)
    print get_shard(1001)
    print get_shard(2000)
    print get_shard(2007)
    print get_shard(4867)