I agree that ELB needs some enhancements. I personally what to find a more open solution then NetScalar, but there is not too much out there.
I dabbled myself with nginx trying to build an auto-scaling LB system, and might resurrect that if the proposal for an nginx based VR/LB goes through. Still, any work which can increase the HA abilities of CloudStacks networking components is essential. I look forward to hearing more on this effort as it grows. Thanks. >>>-----Original Message----- >>>From: Murali Reddy [mailto:murali.re...@citrix.com] >>>Sent: Monday, December 17, 2012 11:58 AM >>>To: cloudstack-dev@incubator.apache.org >>>Subject: [Discuss] EIP and ELB enhancements for HA & Failover application >>>architecture >>> >>>I would like to work on enhancing EIP/ELB functionality present in >>>CloudStack, so that highly available and fault-tolerant application can be >>>architected using CloudStack deployments at a region level. EIP and ELB are >>>both AWS networking features that help building fault-tolerant, highly >>>available application architectures on top of AWS [1],[2],[3]. Idea of this >>>enhancement is to leverage the RHI (Route Health Injection) and GSLB >>>functionalities available in application delivery controllers that provide >>>HA/DR solutions in Active-Active data centre configuration to provide AWS >>>style EIP and ELB functionality in CloudStack. I opened enhancement >>>requests 652, 653 for tracking. >>> >>>CLOUDSTACK-652: High Availability: EIP enhancements >>>CLOUDSTACK-653: High Availability: implement GSLB (Global Server Load >>>Balancing) capability for ELB service >>> >>>Also another complementery effort I would like to work on is close the gap >>>between CloudStack implementation of ELB with that of AWS w.r.t to >>>CloudStack ability to auto-scale up/down request handling capacity. This >>>would require CloudStack to orchestrate provisioning load balancer >>>appliances. I opened enhancement requests 654, 655 for tracking. 655 is >>>pre-requisite for 654. >>> >>>CLOUDSTACK-654: ELB: auto-scale request handling capacity by >>>provisioningLB appliances >>>CLOUDSTACK-655: framework for CloudStack to orchestrate virtual >>>network appliances to provide network services >>> >>>This effort in general would involve two parts. First, coming up with >>>framework/abstraction/generic configuration etc with out any particular >>>assumption of ADC or network appliance. Second, ADC/appliance specific >>>implementation that would realise the functionality. For the second part, I >>>would be primarily working on NetScaler ADC. If any one interested in >>>contributing support for other appliances/ADC I would be happy to work >>>toward integration. >>> >>>I would be starting POC to uncover the issue in supporting cross zone >>>operation with CloudStack. I hope to come with up a functional >>>requirements spec by early next week. >>> >>>[1] http://media.amazonwebservices.com/AWS_Disaster_Recovery.pdf >>>[2] >>>http://media.amazonwebservices.com/AWS_Building_Fault_Tolerant_Appli >>>cations >>>.pdf >>>[3] >>>http://support.rightscale.com/09-Clouds/AWS/02- >>>Amazon_EC2/Designing_Failove >>>r_Architectures_on_EC2/00- >>>Best_Practices_for_using_Elastic_IPs_(EIP)_and_Av >>>ailability_Zones >>> >>> >>>-Murali
