This only seems like a good idea for larger organizations. The idea would be to share a pool of server resources among many working groups with assumptions:
1. server utilization on average would be good
2. seldom have the situation where everyone wants to do large runs at once
At CompassLabs we use Elastic MapReduce: great because we can use a large number of servers for a few hours, and the price is right. On the other hand, large companies like Yahoo run their own large Hadoop clusters (just as an example)
So, this seems like a good idea only for large organizations.
1. server utilization on average would be good
2. seldom have the situation where everyone wants to do large runs at once
At CompassLabs we use Elastic MapReduce: great because we can use a large number of servers for a few hours, and the price is right. On the other hand, large companies like Yahoo run their own large Hadoop clusters (just as an example)
So, this seems like a good idea only for large organizations.