A single server is a multi-system and multi-tenant macro system, then the entire data center or multiple data centers are connected into a super-large cluster consisting of hundreds of thousands or even millions of servers; the system running on it will completely change. It has become a super-large multi-macro system that runs in parallel with multiple cluster systems of different tenants mixed and crossed together.
Consistency of physical hardware and diversity of virtual “hardware”
Many hardware students, involuntarily, would like to engage in hardware “innovation”. for example:
Through PCIe Switch, connect CPUs, GPUs, storage disks, network cards and other I/O devices of different numbers and specifications to form various types of hardware servers with different shapes;
Through powerful capabilities such as smart network cards and DPUs, the free combination of multiple computing nodes with different specifications and different performances can be realized;
In addition, through the enhancement of TOR functions, many DPU work is placed on the TOR switch to realize the way of Smart TOR or TOR-DPU to achieve the enhancement of system capabilities;
Various other hardware innovation projects.