Speech from Raul Coria, IPM, Next Generation Data Centers

On March 24th, Raul, a virtualization specialist at VMware, talked about the new generation data centers and how they differ from the ones we can currently find in companies.
At the beginning of the presentation we saw how the first data centers were designed. Those centers had one PC for each service, therefore, if we are talking about a large company, very large rooms were needed to house all that computing. At that time there was hardly any virtualization, tying the service to the physical PC, causing a small failure in any of these machines to directly affect the service.
 
As the years went by and technology advanced, data centers began to migrate towards virtualization. This technology allows software to be decoupled from hardware, making it possible to run multiple operating systems on the same physical machine. Thanks to this advance, resources can be better optimized and the critical services of a company can be better protected. Still, the networking system (SANs and LANs) is complex and needed to be simplified in order to improve virtualization and reduce the number of cables crossing between the data center. Storage arrays were introduced, where each cabinet brings together all the storage disks and therefore all the data. This technology makes it possible to create redundancies (with the RAID concept) and ease of access to data from any device. The Blades chassis performs the same function as the storage cabinet, but instead of accumulating storage, it stores the computation, having multiple servers in the same place. Finally, devices such as unified switches allow the two types of networks (SANs and LANs) to be linked so that you don't have to use two different devices and thus have greater control over the operation.
 
The presentation focused on how a next-generation data center can take advantage of virtualization to optimize resources and improve performance through redundancy.
To begin with, redundancy in the physical data center is necessary for high availability, since in the event of the entire center going down, it is still necessary for services to be active. From this need come technologies to synchronize data centers with each other, such as wan optimization to improve connectivity or duplicate backup to save data in the event of a ransomware attack.
 
Along with virtualization come a series of concepts that will allow us to have a redundancy of all the elements and an optimization impossible to achieve without this technology.

Vmotion is a VMware concept that allows us to move active virtual machines from one physical server to another, thanks to this we can avoid crashes if we detect any hardware failure, since we will move the service to another functional hardware to be able to check the one that is giving problems. There is the same technology for storage called "Storage Vmotion", allowing to move the storage currently active.
Through the DRS (dynamic resource scheduler) we can balance the resources allowing us to move the virtual machines between the hardware in case their consumption is excessive. Together with the DRS we find the DPM technology (distributed power manager) allowing us to put in standby state the machines that at that moment do not have any virtual machine running, so we will obtain a significant saving in the use of electricity. With the "virtualization fault tolerance" we will be able to have a virtual machine working and an identical copy in standby so that in case the main one goes down, the secondary one is activated making the service they are giving visibly uninterrupted. With the "scale up application" we will be able to add RAM memory physically to an application without interrupting the service and with the "virtual networks" we will be able to create virtual networks within the hypervisor, simplifying the physical network map.

To finish with the presentation, Raul talked about the revolutionary concept of "software defined datacenter" where the software of the physical machines is virtualized allowing each of them to perform different functionalities to those that are designed for, for example, we can have small firewalls for each of these machines installed in the software itself, we can also have switch and routing functions. The goal of this technology is to minimize the use of hardware as much as possible to simplify the data center of the future.