Ron Gordon
Director – Power Systems
High availability is key to any critical solution today, and in many cases, a regulatory requirement. IBM Power Systems has PowerHA System Mirror, as a widely-used high availability solution, currently used by most Power Systems customers. In some cases, the PowerHA System Mirror solution is a combination of over complexity and higher cost than the solution requires. An alternative is now available. It is called Geographical Dispersed Recovery for Power Systems (GDR for short), which has been in use on System z for quite a while, known as GDPS. GDR is basically a VM restart model.
GDR is comprised of a service implementation offering and licensed code called KSYS (for Control System… with a “K”). KSYS runs on a failover, or secondary system, and monitors the primary system VMs that are to be made highly available. KSYS is connected to the HMCs of each system and is also aware of the VIOS’s on each of the systems. On the failover system, no VMs nor software need to be running. KSYS monitors for a “VM failure.” In the event of a failure, an alert or notification is made to an administrator to perform the recovery to the secondary system. KSYS (based on HMC and VIOS information) then starts a VM on the secondary system, connects the network and storage, and restarts the application in the new VM.
If we compare this to PowerHA System Mirror, the RTO may be slower, but there are several benefits to potentially offset this, if near active-active RTO is not a requirement. First, the cost is less since there is no need for any software licensing on the secondary system of the failing application, as in HA. The GDR setup is a simple 10 step process. In many cases, active-active is not necessary. Cost of GDR is low. There are other benefits as well, such as GDR fully supporting Linux on Power while PowerHA System Mirror does not.
Other points that are important to know are: GDR supports Power7 and Power8. GDR can support Enterprise Pools. GDR supports AIX, IBM i, and Linux on Power. KSYS can be scripted to change resources on restart. GDR supports disk replication for the data consistency using EMC SRDF, and now supports IBM Storwize SVC synch, asynch, as well as the IBM DS8000.
This is a rather short description of the new offering, and I am sure you have other questions.
Please contact your Mainline Account Executive directly, or click here to contact us with any questions.