As an industry, we have been looking at cloud-based technologies both from private and public structure and how best to optimize design, engineer and develop such technologies to better optimize the world of wireless and the Internet of Everything.
But one aspect that has not been discussed at length is how poorly hardware and software perform in cloud-based environments. I want to discuss some of the challenges facing the industry and some potential solutions that can help create and bring a new revolution to the world of Wide Area Networks (WAN), along with the automation of practically every human-to-human and human-to-machine interface.
Currently, there are two technologies being discussed in almost every seminar or white paper being published--software defined networking (SDN) and network function virtualization (NFV). While these vary in structure by different vendors, clearly, all of them attack certain aspects of the mobile carrier network or Tier 1 landline networks. Let me give you my two-cents on what these technologies must address:
- SDN must create a more agile network with the development of an open northbound interface. This becomes an enabler for service providers (SPs) to reduce time-to-market for service introduction, reduce capex unit cost by focusing network elements (NEs) to just move traffic, and reducing opex unit cost for network services that take significant human capital cost to deliver, such as establishing protection and restoration or provisioning new connectivity services.
- NFV must enable SPs to provide new services, and hence, new incremental revenue, by replacing dedicated hardware/software located on the customer premise, e.g., DVR, storage, firewall and others.
Cloud computing, on the other hand, must enable enterprises to leverage shared and scalable computing resources, hardware and software to impact their capex and opex unit costs.
The promise of such technologies has always been for
These promises are expected to deliver much better total cost of ownership (TCO) with lower opex and in essence support moving to a hardware-agnostic or independent model, offering further savings.
About a decade ago, I predicted that the battleground in the 21stA century would be all about software and not hardware. Although hardware is needed, it is the role of software to optimize all five functions above using new state-of-the-art technologies such as SDN and NFV.
The problem that can become very complicated is that enterprise customers' networks and appliances are not designed for multiple tenants, pay-for-play or on-demand services. However, SDN and NFV are fundamentally designed for such functions. That means that it is imperative for CXOs to sponsor corporate-wide programs to move into SDN and NFV, offering capabilities to drive higher revenues while competing for device replacements at the network margins from mobile access points up to wireline or Wide Area Networks.
SDN, by itself, is not really a new technology and has been in existence since 2006. It has been used to mainly improve data center performance, since the concept of big central offices with large Class 4/5 switches are pretty much obsolete in the 21stA century. A
But SDN has a long way to go to deliver an agile network. Today's management of transport networks does not match the agility of the cloud-based services being deployed on them. These two have to converge to bring the transport agility into the 21stA century for service delivery. Why should it take weeks and months to establish a new enterprise customer on an SP network? Why should it take weeks to provision high-speed point-to-point connectivity with specific protection requirements? SDN has yet to deliver just that.
NFV, in contrast, was introduced between 2010 and 2012 to operators in order to improve service time-to-market and network flexibility and allow a smooth transition to the cloud with significantly lower opex. In my view, the sky is the limit on NFV. For any onsite services (e.g., storage, firewall and DVR), whether today or in the future, NFV gives SPs the opportunity to deliver both consumers and enterprises major benefits, such as having a turn-key solution that lowers costs and improves quality of service (QoS).
The initial applications of SDN and NFV have changed greatly over the past few years. SDN focused mainly on cloud orchestration and networking, while NFV focused on IP-based protocols and capabilities such as DNS, DHCP, DPI, firewalls, gateways, and traffic management.
From my perspective, I believe NFV has already taken over Layer 4-7 of the SDN movement by delivering lower capex and cycle time, creating a competitive supply of innovative applications by third parties and introducing control abstractions to foster innovations that carriers need in order to compete with all over-the-top players.
Let's also note that the new world requires openness in almost every API layer of the network from access to the core. The issue is legacy systems and processes that need to be changed in order to adapt to the new world of SDN and NFV.
Nowhere is this more critical than the mobile and Tier 1 landline carriers.
In essence, these sectors need to change all analog processes using legacy systems into digital processes, in which NFV can easily fit. That transition may take years, if not a decade, before it is fully implemented.
But the question is whether MNOs and Tier 1 carriers can wait that long to implement NFV and get the most optimized set of solutions in order to compete globally.
My guess is no, they cannot wait and stay competitive. The transition to NFV can be done more quickly, and I'm going to tell you how. A
Part 2 coming soon...
This story, "Impact of Today's Hardware and Software Applications in Cloud-based Environments: Part 1" was originally published by Network World.