In the last 20 years we have mastered building and operating the infrastructure, now we at IT are broker of services. You are managing the service between Microsoft or Amazon and end users.
On premise we scale-up, on the public cloud scale-out. On premises we assumed a reliable infrastructure (we designed, configured, manage it this way), on the cloud we should expect infrastructure failure. In service oriented architecture (SOA) the application is aware of the infrastructure, knows how to reconfigure or recalibrate (ex. If it requires more databases- it can make database services requests). Traditionally on the premise the application is completely unaware of the infrastructure. In order the service to be service oriented and the application to be designed for the public cloud, it needs to be written that way -the application to manage them and understands infrastructure.
On premise is fixed cost, buying upfront. With the public cloud is usage-based cost. On the cloud there is a lot of automation of the management of the instances.
Are there really cost savings OpE vs CapEx? With on premise servers we are used to overprovision. If we learn the differences between the public cloud and the on premise- we can save money.
The flexibility and elasticity of the Cloud is very helpful- you can change hardware at a reboot.
Security and Compliance
If you don’t secure the public cloud is your fault. You are provided but you have to implement it.
Currently there are around 94% machines on the premise vs. 6% on the public cloud.
Having servers on premise and in the public cloud (the hybrid model) is most likely the future. The cost for the public cloud is continuously going down. The public cloud will be the default and the exception will be the on premise.