Hyper-converged systems are rapidly increasing in popularity due to the tight integration of compute, storage and...
networking, and the resulting ease of deployment and management.
While tight integration is desirable, these systems have many of the same networking challenges as other data center deployments, including requirements for scalability, automation, security and management of traffic flows. Additionally, they need to link to other data center resources inside the data center, at remote data centers and in the cloud. Software-defined networking architecture can ease some of the scaling, automation, security and connectivity challenges of hyper-converged system deployments.
Hyper-converged systems integrate storage, computing and networking into a single system -- a box or pod -- in order to reduce data center complexity and ease deployment challenges associated with traditional data center architectures. A hyper-converged system comprises a hypervisor, software-defined storage and internal networking, all of which are managed as a single entity. Multiple pods can be networked together to create pools of shared compute and storage.
The hyper-converged market is expected to grow rapidly in the next few years, from approximately $1 billion to more than $5 billion. Hyper-converged system providers -- including Nutanix, Dell EMC and VMware, and Hewlett Packard Enterprise (HPE) -- have focused on integrating compute and storage into a single, scalable resource.
Networking requirements for hyper-converged systems
Networking in hyper-converged systems is dedicated to establishing and maintaining communications between the compute and storage elements and is typically proprietary. Some key networking elements required for hyper-converged systems include the following:
- internal data center networking, like scaling hyper-converged system pods or connecting two heterogeneous elements -- e.g., connecting HPE and Dell compute components;
- data-center-to-data-center connectivity via the WAN;
- data-center-to-cloud connectivity; and
- remote access from branch and other locations.
Where software-defined networking architecture comes in
In software-defined networking architecture, the network control plane is abstracted -- or separated -- from the underlying network hardware. Along with this abstraction, software-defined networking contains the following attributes:
- automated deployment, configuration and management, support for multi-tenancy, and ease of scalability using cloud resources;
- easily adaptable, customizable and programmable with open APIs; and
- open, standards-based, multivendor and interoperable with a variety of network software.
Numerous aspects of software-defined networking architecture relate to hyper-converged systems in a data center. For example, it can help connect or manage internal traffic flows, data center communications and networking outside the data center.
Network vendors, such as Big Switch Networks and Cumulus Networks, currently provide elements of software-defined networking architecture to help customers scale and network hyper-converged system clusters. Other suppliers -- including Cisco, HPE and VMware -- are enabling communications between hyper-converged systems and other data center and cloud-based elements.
Dig Deeper on Software-defined networking
Related Q&A from Lee Doyle
SD-WAN technology is available in a variety of business models, including SD-WAN as a service and managed SD-WAN. But how do the different ... Continue Reading
Make the case for an SD-WAN implementation, and explore the benefits and main use cases for SD-WAN in enterprises, beyond traditional hybrid WANs. Continue Reading
When contemplating SD-WAN managed services, organizations should ask questions about network security, infrastructure compatibility and bandwidth ... Continue Reading