Understanding Deployment Strategies
The deployment strategy of a service determines how new versions will be initialized. Liferay PaaS provides two deployment strategies for its services: RollingUpdate and Recreate.
Differences Between Deployment Strategies
Users expect applications to be available all the time and developers are expected to deploy new versions of them several times a day. In Liferay PaaS this is done with the RollingUpdate strategy. Rolling updates allow deployments to take place with zero downtime by incrementally updating instances with new ones as long as the cluster has at least two working nodes.
| Deployment Strategy | Pros | Cons |
|---|---|---|
RollingUpdate | New versions are released incrementally, maximizing uptime | Full roll outs can take additional time relative to number of instances |
Recreate | Application state entirely renewed at once | Downtime length that occurs is dependent on shutdown and boot up duration for the application |
Deployments defined with the Recreate strategy terminate all the running instances before recreating them with the newer version.
In general, the RollingUpdate strategy is our recommended approach. The Recreate strategy can be used in non-production environments or in some cases where it might be required (e.g., if you’re updating to a version of Liferay DXP that changes the database schema).
Configuring Deployment Strategies
By default, the services in Liferay PaaS are pre-configured to use the RollingUpdate strategy. The deployment strategy for each service is configurable via the strategy variable in the service’s LCP.json:
{
"id": "myservice",
"strategy": {
"type": "Recreate"
}
}