Now that we have an idea of where we are starting from, and have started to look at where we want to go and why, let’s look at some prerequisites.
Service consumption is based on two key concepts:
Hardware virtualisation
With hardware virtualisation, the underlying hardware, CPU, memory, and storage, is virtualised into a hypervisor that standardises how the server operating systems (OS) see the hardware. This is a key step - without virtualisation, migration to a cloud becomes a completely new build of all IT systems from scratch. Virtualisation on the owned IT system can greatly accelerate the move to the Cloud.
Hardware standardisation
Virtualisation will help drive standardisation of the hardware, moving away from bespoke hardware solutions unless a specific business outcome is identified. But standardisation needs to begin moving up the IT stack to take full advantage of the cost and agility benefits of Cloud Services. This includes looking at operating systems and application platforms to reduce them to core business outcome needs.
Full standardisation is not required to consume Cloud but the process needs to begin. Unsupported OS and applications that cannot be run on virtualized hardware need to be prioritised for evaluation to retire or upgrade to a newer, often times supported version.
Example:
“We are a retail store for a non-technology product. We do not manufacture the products and we do not develop them, just sell them in a retail shop. So why do we have so much technology?”
Here we have a massive server/database/application bespoke data centre with hundreds of servers running thousands of applications:
10 different hardware platforms
32 different operating systems
6 different databases running 10 different versions
1000+ different applications, most of which we did not develop, just purchased, not all of them are supported anymore
10+ different java/.NET versions and clients
200 different web portals and thin and thick client applications
Virtualisation
As part of the virtualisation process we need to ask ourselves some questions:
How did we get here? Often the answer will highlight a technology-centric view of your work rather than a business outcome view.
Who owns the delivery of technology? Ownership implies responsibility and often leads to conflicting responsibilities not aligned to the ultimate business strategy and outcomes.
First, we pick up as many of the servers as possible and virtualise them onto a standard hypervisor. The hypervisor of choice is less important than the cost and supportability of the hypervisor on existing hardware but if the plan is to move to Cloud services then x86 hypervisors MUST be preferred.
Available hypervisors on x86:
VMware
Xen
KVM
Hyper-V
Oracle VM
RHEV-H
This is often already accomplished and many organisations have already made this move to consolidate costs.
Complicated applications
In this process, we can identify several servers that cannot be virtualised or applications no longer under support or that will not run on newer operating systems. This is the complicated ‘20%’ of the 80/20 rule and should be documented, and then analysed, for simple decommissioning or separate projects to virtualise them. These complicated applications should not stop the process of moving to the Cloud. In business transformations, these complicated applications often highlight broken or antiquated business processes that if fixed allow you to decommission them outright.
As servers and applications are virtualised they can begin moving to an IaaS cloud. Often the data, storage or databases may move before the virtualised machines but a Cloud Centre of Excellence can help identify the optimum sequence of migration.