All of the instances in the CirrusGrid environment (like databases, application servers, etc) are isolated containers located on different hosts.
Three main reasons why CirrusGrid provides individual role-based servers are:
Isolated Containers Live Migration #
In some cases as your application starts requesting more and more resources, the physical server running your node might not be able to provide the required resources. In this case, Jelastic can perform a live migration of the node to another host within 30 seconds.
During that migration, the application keeps working being scaled vertically. As an application scales within a server, other applications can be migrated to another server to make room. Live migration allows CirrusGrid to deliver all the needed resources for applications without restarting containers and causing application downtime. In addition, it can be used during maintenance or other planned downtime to perform automated evacuation of containers from a physical server.
Also you can add more instances to your environment using horizontal scaling, not having to worry if there is enough space on the host. CirrusGrid chooses the host with an appropriate amount of free space and relocates your server, in order to offer high-quality performance to each of your nodes.
High Availability for Applications #
CirrusGrid platform provides the highest availability by distributing containers of one environment evenly on different physical servers (Hardware nodes). This is done with the help of anti-affinity groups configured to specify that certain containers should never run on the same physical server.
As a result CirrusGrid eliminates any risk of application downtime if one of the physical servers has any issues with performance.
Security of Isolated Containers #
CirrusGrid uses the virtualization method to run multiple containers simultaneously on a single host. These containers are fully isolated with no risk of interfering with one another. If the security of any container at the hardware node is compromised, the rest of the containers are left unaffected.
Note: Since each instance within Jelastic PaaS represents an isolated container, it can’t be reached from another node with a simple localhost:port_number or 127.0.0.1:port_number reference - the corresponding container hostname or private/public IP address should be used instead (for example, this could be required to interconnect your application with a database instance).
