We regularly do deployments of new functionality and fixes to our customers websites that run on portal. That being, for example, a new Theme, updating EARs and WCM changes.
For some of our customers we like to have zero downtime deployments, as sometimes deploying outside working hours is not an option.
With the traditional portal installation we have done that by splitting the amount of licensed PVU in half to create two zones. Each zone has its own database, IBM HTTP Server and portal cluster.
The traffic coming in is going through a load balancer to the two zones.
Content is pushed from an authoring portal to the two zones.
For deployments we can take a zone out of the load balancer, and deploy on that zone. Then switch the loadbalancer and deploy to the other zone.
This approach comes with a cost as a lot of components are duplicated, like the database, HTTP server, deployment manager and the search server.
In the new docker/kubernetes environment is there a better way to achieve zero downtime deployments? What would the architecture be in that case?
From a license perspective; today the HCL Digital Experience licenses provide permissions to use several components restricted for use with the HCL DX platform operations. The HCL Software License site presents the DX licenses https://www.hcltechsw.com/wps/portal/resources/license-agreements; they are also available from the HCL Software License Portal downloads (Flexnet) for DX entitlements. In addition to the PVU entitlement model now in place we expect to offer additional metrics to more closely track DX container deployment utilizations in the future.
The HCL DX 9.5 Help Center pages focused on Containerization will continue to add guidance for deployment scenarios. As always, contact HCL Support where you may need specific instructions to ensure your deployment practices remain on track.
Hi Lauren,
Basically we are looking at saving infrastructure cost (AWS). So we are looking for a way to minimize duplication and still be able to deploy without an outage. The dependency on external database and file storage is complicating this. We can not find a proper reference in the documentation.
We plan to offer more published guidance to manage container deployments in the Help Center topics. In the Containerization pages, there is a sample CI/CD pipeline script available today: https://help.hcltechsw.com/digital-experience/9.5/containerization/ci_cd.html
Hi Lauren, this tells us nothing about how to prevent an outage when deploying...
We will proceed with a solution where the database server is switched off and a separate zone is spinned on when a deployment is there.
Contact HCL DX support as well where any specific guidance may be needed to ensure best success. Thank you.