Instance auto-scaling: reactive
As a developer or app administrator, I want a provisioning mechanism to specify rulesets/triggers to scale up/down the number of instances according to resource monitoring metrics. This mechanism should have hard boundaries to avoid unbudgeted expenses. [Please comment on what resource metrics matter to you - CPU utilization, bandwidth, # requests, queue length etc.]
Released. See WASABi BETA
http://bit.ly/phMZj9
-
As we are prototyping this, we’d like to know:
- Who’s more likely to own the autoscaling rules? IT pro, dev,
- With proactive rules (defining min/max instance counts for a role) and reactive rules (indicating changes in instance counts based on the values of certain KPIs), would recurrence patterns only need to apply to proactive rules or would it make sense to have recurrence patterns for reactive rules as well?
- Would aggregates of KPIs be needed? E.g. Avg CPU usage in the last 30 minutes. If so, which kinds of aggregates would make sense?
- Would rules involving multiple KPIs be needed?