The dedicated cloud is where by you get the benefits of dedicated servers, such as having all the resources to yourself but with the flexibility of the cloud where by everything is very flexible. You can upgrade and downgrade easily at any time and it ensures high availability at all times.
In the instance where by you have a dedicated server and something goes wrong, there would be some downtime expected until the issue has been resolved. With the dedicated cloud, you would be able to migrate all your services from one host (server) to another host (server). This would be done automatically by the dedicated cloud.
Here we have a cluster which consist of 3 host. The host themselves are built from CPU and RAM. Each host has its allocated resources within. Let’s say we have all our front-end services running on the .50 host and we would use our .51 host for all our back-end services. In the event where by an host is unavailable, the HA (High Availability) would be triggered, the services from either .50 or .51 or both will be migrated to .52. This will ensure our services remain online all the time.
As mentioned previously, the dedicated cloud is very flexible, during the time where by .52 is working both the front-end and back-end, we could temporarily obtain a .53 to ease the load so we can separate our front and back-end.
Once everything is back in order, we would resume all our services back to their default configuration, .50 will be the front-end, .51 will be the back-end and .52 would be our spare, backup or dev environment. This is how the dedicated cloud can be advertised as 100% availability.
HA (High Availability)
HA is a process where by the cloud will do everything it can to ensure the services or in this case, the virtual machines to remain operational at all cost. This may include over committing the host in the cluster to work harder or to take higher loads to ensure the service are active regardless of the situation.
This is feature when your virtual machines exist on both host at the same time. It means that if one of the host experiences issues, the other host will keep the virtual machine running. The down time to this is that for the same virtual machine your consuming double the amount of resources as it has to be reserved on both host machines in the cluster.
DRS (Distributed Resource Scheduler)
With DRS it will optimise all your current virtual machines against your current resources in the cluster. If your cluster has 2 host and there is 2 virtual machines in the same host, DRS will look to move one of the virtual machine to the other host so that the resources are automatically balanced on all available host.
To move services from one host to another typically involves powering down the virtual machine and then to move it to the other host, afterwards you can power up the virtual machine again. This leads to your virtual machine being unavailable during this period. In terms of a web server this isn’t ideal. With the use of vMotion you have the ability to perform a hot migration of services from one host to another without any interruption.