Virtualization allows the creation of a layer of abstraction between services/applications and the physical layer needed to deliver them. This allows each service/application to dynamically access much more resources than the ones available on the physical servers where that service/application is running. This bears also a lot of other advantages in terms of asset consolidation, operational efficiency, business agility and cost control.
Hyperconverged software sits on the top of virtualization layers and proposes an approach where compute, storage and networking can be provided by the same physical device (a server). In a hyperconverged schema, each server adds to global pools of resources which span cluster wide and are accessible to application and services.
This is all good! But it comes with a price to pay, in terms of performance, scalability and efficiency, due to the addition of fat software layers to the infrastructure stack.
The goal of physicalization is to provide the same advantages of virtualization and hyperconvergence, but with much lighter software layers and with hardware taking up part of the intelligence and logic.
Physicalization is implemented by a hardware and software integration that makes it possible to create global pools of physical compute, storage and networking resources, and subsequently allocate them dynamically to virtual machines. This approach enhances the idea of software defined infrastructure into software controlled, hardware accelerated infrastructure. It removes the processing overheads typical in virtualization of I/O and creates a flexible number of physical devices per VM.
In other words, physicalization doesn’t emulate hardware in software, but leaves hardware in hardware, keeping the flexibility of software defining resources.
The net results of physicalization is achieving the same agility and effectiveness of virtualization/hyperconvergence but with more performance, less latency, more scalability and less power consumption.