For the time being, no IT player has a complete SDDC offer (Software-Defined Datacenter). A finding that companies have to do, without their knowledge. To answer this problem, they must assemble a battery of software and hardware, each from different suppliers. While the hyperconverged infrastructures pre-assemble some programmable parts , they can not form a complete SDDC.

Before you buy and implement a complete SDDC, you first need to understand what needs to be met in the enterprise, and which technology may ultimately be suitable.

Kết quả hình ảnh cho tech cloud

How to define an SDDC

A Software-Defined Datacenter is a conceptual infrastructure in which each element is controlled via abstraction layers, automated procedures or a set of rules. Infrastructure as Code is a key component of this programmable platform . Traditional IT infrastructure relies on manual configurations of isolated components. An SDDC, on the other hand, uses source files instead to describe the target infrastructure, favoring automation and producing a consistent and reusable configuration. These capabilities are the foundation needed to deliver the services.

A complete SDDC is almost entirely automated. Developers or business departments "consume" the infrastructure services through APIs (Application Programming Interface) - which promotes automation. Most companies are not ready for this, hampered by business processes and the human factor that slow down the implementation of automated procedures.

However, you do not ultimately need a fully automated infrastructure to benefit from SDDC. The latter includes several programmable components, such as:

Hypervisor: Programmable CPU and RAM with certain storage and network functions;

SDN (Software-Defined Network) for the physical and virtual network;

SDS (Software-Defined Storage) to group or segment storage resources;

Configuration management tools for hosts, virtualized OS, VMs and applications;

Software such as Docker or other container management solutions;

Operations, such as backup, disaster recovery, capacity management and performance.

Instead of managing each dimension of the SDDC directly by manipulating the elements of each VM, application or server, the IT administers an SDDC via a list of rules. To manage automation, each programmable element must have an API . With programmable platforms, policies are applied programmatically to things like VMs, compute, network, and data center storage.

The role of hyperconvergence in the SDDC

Hyperconverged infrastructures are initially composed of programmable compute, associated with SDS. All solution providers offer a form of distributed storage over a hypervisor and include provisioning and management of physical servers. The hypervisor brings the programmable network dimension to each of the physical nodes. It's a start, but it's far from complete.

One of the SDDC engines is the API for configuration and provisioning. In the case of a full SDDC, a hyperconverged appliance must allow the automated deployment of additional nodes and this form of automation must be versioned. Ideally, the implementation of dedicated policies should help to bring together or automate the allocation of free capabilities.

Some hyperconvergence providers go further in the SDDC, for example with backup and replication features, which are built into the storage. In a hyperconverged infrastructure, backup and replication must be controlled by policies aligned with business needs. For example, a policy might specify that a backup of a critical system is performed every 4 hours. Another copy of the same system is never done more than one hour after the production version - in a disaster recovery scenario. This approach is key for an SDDC and must be associated with capacity provisioning.

Most solutions on the market do not offer SDNs to interconnect nodes and VMs. None also integrates the configuration of the physical switch. Instead, they rely on the hypervisor for virtual network configuration. None of the hyperconverged platforms manages the OS or applications in the VM; some tools need to be added, such as configuration management tools like Puppet, Chef or Ansible.

If hyperconverged solutions do not offer a complete experience of SDDC, it is because it is not their primary role. On the other hand, they form an adequate platform on which to build a SDDC.