I have spoken to customers and also spoken to some colleagues and realize the IT paradigm as changed quite substantial. However many are still operating the wrong way which end up even with virtualization, cloud or Software Defined Data Center, etc. it is still not helping them achieve in a big scale.
Recently I did a session on BC/DR: Implementing a Holistic Strategy with VMware Solutions session at APJ vForum in Singapore. I mentioned a few things that was abstracted from VCDX Design Defense Blueprint page 11. This diagram has helped me in a lot of discussion and I see many are still not aware. The five things: Availability, Recovery, Security, Manageability, Performance.
Taking an example today, when a Application owner request a machine with certain specification in terms of CPU, Memory and space from an Infra personnel, the answer was: "Just provision as requested". In this scenario, it doesn't matter if this is physical or virtual, if the machine was to have issues or went down, who is always held responsible? Infra guys. Have you ever thought that both are at fault? First did the Application owner knows what he really wants? Did the infra personnel asked him the details? Could think be different? Would things like availability requirements, RTO, RPO, security, etc come into play?
In bigger organization, many would think that they have the budget, they could implement lots of solutions, automate them but in a smaller enterprise or businesses we cannot do that. Is that really true? Even in this big organization, do you think the infra personnel asked those questions? They might not but why they ain't they facing the same in the smaller enterprise?
Let's go back to my session on BC/DR. When application owner request a machine, questions like what is your RPO, RTO and availability could be asked? So if an organization has a daily backup (assuming they do daily backup for everything unless stated otherwise) they will place that machine into the usual job schedule. This will end up your RPO to be 24 hours (once a day equate 24 hours). If this is a database, would this be sufficient? Or do you need a RPO of 1 hour? In terms of availability, do you have a SLA? Any clustering requirements or application redundancy solution? vSphere has App HA, physical world can leverage on Veritas cluster or MSCS. Was this mentioned and questions? Say the application owner has $20,000 to spend on this machine, but requested a five 9's SLA, can this be even achieved? If not, what is possible or alternative? Today, the IT Infra personnel is no longer just someone who administrate the environment. He act as a consultant to the company and he has to questions and provide advice. His responsibility has just grown bigger. By missing this out, the expectation from both parties will be different. Technology has advanced, human thinking need to be enhanced.
Now let's goes back to the normal environment on request. Same thing would have applied. If an application owner request for a machine, would you check what is their requirements? In bigger organization, why do they do so well in term of the process? Is it not the technology? That is partially correct. The consulting creating this process was either correct or the process all along in the organization was set correctly. So in a big organization today, when someone requested for a machine, all this questions would be ask and chosen by the requester when they submit the request. E.g. A Dev machine could be auto assign to a Tier 3 storage where functional test is expected not a performance test. Production machine is automatic place on a Tier 1 since they would need that performance. They could also be asked, if Availability is checked, MSCS will be implemented else the standard vSphere HA is used instead.
As people start to use these systems, the consultant role mentioned previously from the infra personnel start to goes away. In such, big organization might still have the the usual IT personnel who didn't have the knowledge cause the system has already done the work. What has changed, a proper workflow was implemented to automate the process. What is different here was the process. Can this process be used in other places? Answer is Yes!
Back to organization that cannot automate or pay for big consultation fee to do that, can they provide a better service up front? Can they enable the process to be more efficient? Can application owner be realistic and co-own responsibility? The answer is Yes! So would automation tool be useful in an Data Center? Yes if you got the right process implemented. No, if you don't if that automation tool is just a tool without a proper process first in place.
How many times you hear people introducing different automation tools or solutions and tell you process need to be in place? How can you help you if that?
People makes the different and I see many professionals are still holding on to old IT paradigm and this has to change when technology evolved, the mindset has to shift. Do something different today, questioned what is required? What is the alternative? No one knows the environment better than the infra personnel. Not only do you value add to the company, you also create a niche for yourself. You also get both parties have the same expectations and during an issue the responsibility of the machine is not just you alone.
Sit down. Have a thought. By asking shifting your thinking can create a whole different, will you?
This is strictly a personal point of view. There can be many others reason and debates. In summary, the IT Paradigm is no longer the same and the way we think and do should also reflect the changes.