1 person found this helpful
Normally patching would be done at a specified window time and would require a downtime. A list of available patches would be presented to the customer and eventually its left to the customer to decide which patches to install.
But a backup of the domain will happen automatically and subsequent to that patches would be applied. (Restart of managed servers happens ). For security patches, Oracle will notify all the customers at a fixed maintenance time.
Instances processing during patching time would always depend on below
1) how patching will be implemented.
2) how the infrastructure is setup.
If the infrastructure has load balancer VMs, then you can control the traffic going to individual managed servers ( either by using some sort of Heart Beat services deployed on managed servers or edit the load balancer config manually ) and once all the running instances on the managed server are completed successfully, then patch the individual managed server.
If load balancer VMs are not available, you could stop the individual managed servers smoothly and then patch them individually.
Although some downtime is expected during an upgrade cycle. Hope it helps.
Thanks for the inputs.
Since cloud hosts multiple customers, generally upgrades happen together for them and was thinking it cannot be at the control of the customer. More specifically, I was worried about major upgrades ex- 11g to 12c (though this may not apply to the present cloud - but would apply to the next new SOA release) than minor patches. I am ok with downtimes - but more concerned about the in progress composite instances and how they will work post upgrade. Any guidelines around that?
1 person found this helpful
The below links could be helpful for you.
https://cloud.oracle.com/SOA?tabname=LearnMoreInfo&lmResID=1439473853567 ( How does patching and upgrade work for SOA cloud service )
https://docs.oracle.com/cloud/latest/jcs_gs/JSCUG/GUID-A0491DD0-6E6C-405D-A00C-A213AABDF91C.htm#JSCUG-GUID-B35AE4EE-14DF… ( Start/Stop managed servers & load balancer VMs )
I notice that this is the second place where it is mentioned that SOA Cloud Service is for DEV/Test Environments. Is it not production ready?
What use cases are supported by SOA Cloud Service?
There are 4 primary use cases:
- API Virtualization – Using API Manager Cloud Service to virtualize your back end services, and publish them to a cloud portal where your developer community can discover them.
- Dev/Test – Using SOA or Service Bus in the cloud to quickly set up dev and test environments while still running SOA or Service Bus on premise for production.
- Lift-n-Shift All Workloads – Run existing SOA or Service Bus projects for Dev, Test and Production in the cloud.
- Born in the Cloud SOA – New integration projects can be started with Development in the cloud, move to Test in the cloud, and finally to production in the cloud.
The above statement reflects various implementation scenarios how one could use SOA cloud service.
Point 2 above mentions about how quick the turn around time would be for developers to hit the ball running from start without one has to wait for long in setting up DEV/TEST environments in on premises using traditional procedures. We know how much time it takes to set up a minimum feasible DEV/TEST environments.
So ideally using SOA cloud service one can easily spawn a VM on cloud for DEV/TEST purposes. It is production ready and if you have an Oracle Support account, then suggest raise a ticket to get more information about the same. Hope it helps.