CareerCup

Can you clarify the problem?

Can you explain the problem little bit more. It seems that services are like a circular queue, and the order are put in one by one. now when m>n then you dont have services left to process the order. It can be solved if you dynamically add few more services.
Thanks,
DKJHA

No. These are just service instances. Each represents an instance of the same service... same program, various running instances.
so orders are all stored in the database
and each service handles only those orders which hash to it by the rule:
orderid % m + 1 = service id.

load balancing algorithm

If the order ids are random integer, then it's likely at a particular time some of the service s_i may have no order to process, whereas some service s_j has more than 1 orders to process. So this is some balancing issue. A simple approach could be counting each services status dynamically. The given static approach takes O(1) time for assignment with balancing problem, dynamic approach would take O(n) time which may alleviate this problem.

Any useful link/tutorial for load balancing issues? Wiki entry doesn't have much details.

search for "persistent hashing"

if m<n then service 1 will never be used.So i think this is the problem.

if N is greater than M, then there might be a chance that the order is not assigned to the service. I think may be the question is put incorrectly. I think it should be S_n % j should go to Oj. I am taking it as a Hashing problem.