In part 5 the paper tells about the replica selection for subqueries, it says that if all hosts are similarly loaded, they prefer nearby replicas and if they are at similar distance they prefer the most lightly-loaded host. Is the priority here given to the distance or the load? For example if there is a replica that is further than the others but has a light-load may have a small response time, but how do they assign it?
In part 5 the paper tells about the replica selection for subqueries, it says that if all hosts are similarly loaded, they prefer nearby replicas and if they are at similar distance they prefer the most lightly-loaded host. Is the priority here given to the distance or the load? For example if there is a replica that is further than the others but has a light-load may have a small response time, but how do they assign it?
ReplyDeletenevermind... I now see that the formula given includes the distances while calculating the response times
ReplyDelete