amazon web services - Using AWS Autoscaling as dispatcher -
i have been given business logic of
- a customer makes request services through third party gateway gui ec2 instance
- processing time (15hr)
- data retrieval
currently implemented statically giving each user ec2 instance use handle requests. (this instance creates sub instances parallel process data).
what should happen each request, ec2 instance fired off automatically.
in long term, thinking should done using swf (given use of sub processes), however, wondered if quick , dirty solution, using autoscaling correct settings worthwhile pursuing.
any thoughts?
you can "trick" autoscalling spin instances based on metrics: http://docs.aws.amazon.com/autoscaling/latest/developerguide/policy_creating.html
so on each request, keep track/increments metric. decrement metric when process completes. drive autoscalling group on metric.
use step adjustments control number of instances: http://docs.aws.amazon.com/autoscaling/latest/developerguide/as-scale-based-on-demand.html#as-scaling-types
interesting challenges: binding customers specific ec2 instances. have hard requirement of giving each customer own instance? sounds autos calling better suited paralleling process of actual data, not requests routing. may away having fixed number of machines and/or scale based on traffic, not number of customers.
Comments
Post a Comment