This post is about my experience over the past 10 years of promoting and progressing virtualisation and cloud automation within the Middle East (Saudi Arabia in particular). However, people are people, wherever you are on this spinning globe we call Earth, these lessons apply everywhere.
The lessons learned:
- Full CIO support – without it, you are doomed to stall and fail.
- Technical Champion – the guy who puts his hand up to own the project and make it happen. If you are reading this post, that person is probably you.
- Approved budget – if not, take the time to develop an extensive Business Case (including H/W, S/W, PSO, RE, Training, Hiring and Support) for the entire strategy highlighting the CAPEX and OPEX savings. Do not start without the cash to make it happen; you want your project to execute in a timely fashion.
- Training program for existing staff – spread the word and get everyone on board. You will have a lot of staff who do not understand virtualisation (but think they do) and will resist it due to half-baked ideas and misconceptions – Educate them.
- Recruit new staff (Permanent-hire) who have the necessary skills to complement your existing team.
- Use Professional Services (project-based) and Resident Engineers (operations-based) for the first few years. Reduce your operational risk by employing the services of experts: plan for success and knowledge transfer.
- Approved Enterprise IT Architecture – if you do not have this, get one. Follow this action plan.
- VCDX-level design that complies with your Enterprise IT Architecture, pay special attention to Disaster Recovery.
- Phased approach – Rome was not built in a day! Technically you could implement cloud automation on Day 1, however it is going to take time for your people to catch up and get used to a new operational paradigm. Suggested project phases:
- P2V existing physical servers and all future servers will be virtual
- Advanced operations and management
- Cloud Automation/Hybrid Cloud
- Testing: Functional, Acceptance, Performance and Recovery – make sure you fully complete the test cycle. There will be problems, catch them before you go “Operational”, migrate your Development and Test environments first.
- Before each phase of the solution is migrated to Production, ensure that the Standard Operational Procedures exist and are fully tested and owned by “IT Operations”.
There is a common misconception within the industry that the expense of the solution is directly correlated to the technical superiority of the solution. NOT TRUE. A properly designed solution that meets your business requirements does not have to be prohibitively expensive.
This post is also applicable if you are planning Application transformation within your Data Center: eg. HP-UX/zOS/AIX applications to Red Hat Linux on Intel with vSphere. It just means that the project is so much more sensitive to latency and delay due to every silo within your organisation being impacted. Tread carefully!