12-18-2014 02:19 PM
We are using SteelApp to manage the autoscaling of our AWS based system i.e. the SteelApp Traffic Manager, not an external system manages this. The problem we are facing is that regradless of the type of monitoring we configure, the SteelApp Traffic Manager add nodes to the pool before the AWS instances have completed their status checks. This results in client traffic being sent to nodes which are not in any way ready to accept the request. The end user is then presented with a service not available message when in reality, the system is just scaling out. We are planning to prescale our system prior to load arriving (we would have to do this anyway to some extent). However if we needed to add more nodes during peak load or relied on autoscaling, our end users would still be presented with an unreliable service. I feel the lack of effective autoscaling in the (AWS) cloud is a major shortcoming of this software as it does not support the advertised functionality. I have also raised this as a support issue. I'm told it is a known bug but there are no plans to fix it (Zeus bug 24145). If this is true surely it is impacting a large customer base. Do I have bad information? Has anyone else had this issue?
Solved! Go to Solution.
12-21-2014 10:46 PM
Just updating this thread for the sake of completeness. We have been working with Chris in the background through the Riverbed TAC to provide a solution to this problem.
If anyone else is hitting this issue, I ask you to please create a support case with the Riverbed TAC (remembering to set the appropriate priority on the case to reflect the impact on your business).