Many companies are evaluating cloud-based infrastructure services to be the primary hosting location for their enterprise applications, i.e. the software that runs their business. This could be email, CRM, finance packages or any of a host of other enterprise software products. The business drivers behind this decision are discussed elsewhere. Here, we will look at a typical scenario to explain how the cloud service is assembled.
The diagram below is a logical breakdown of some typical elements of an enterprise email solution for a mid-sized organisation:
This would be for either a virtual private or private cloud deployment. In consultation with the managed hosting provider, the ideal design would be agreed. In the above, there are three transport servers, three mailbox servers, a directory server, a mobility server (e.g. to provide email on smartphones, tablets or via browsers) and an anti-spam/virus server. In the pre-virtualisation, pre-cloud era, these would have required nine physical servers hosted on-premise in a server room. Now, the customer pays monthly for nine virtual machines hosted centrally (usually with managed backup to another site) on a provider’s cloud.
For a virtual private cloud, where the infrastructure is multi-tenanted, this would be all the involvement the customer would need. The hypervisor functionality on the cloud platform would “spread” these nine VMs across multiple physical servers, with high availability mode ensuring they will not be “lost” in the event of a physical outage.
For a private cloud, the customer and cloud provider would agree on the physical configuration too. All nine VMs could exist on a single blade server or the customer could pay for additional hardware to position the VMs across two or three (or more) physical servers for added resilience. Once agreed, the cloud provider would manage the deployment going forward, including backup, disaster recovery, and other value-adding services.