Is the Internet breaking?

Wednesday July 14, 2010



REACHING THE LIMIT: The cost of building the Net is increasing substancially compared to revenue obtained from Internet services -- and this will lead to lesser profits to reinvest in innovation in the future. - AP

By WAN AHMAD KAMAL

THE Internet has become part of our lives faster than any technology in world history. So it's fair to ask - how's it doing? Could it be breaking?

Well, it's certainly creaking! From the first web browsers in the early 1990s to today's 1.7 billion users - the Internet has grown exponentially.

One petabyte can hold about 20 million, four-door filing cabinets full of text - or 500 billion pages of printed text. Using a conservative methodology, growth in petabytes per month will result in Internet traffic growing almost 30-fold by 2020.

While growth is good and drives us forward, and making new things possible, the economics of the Internet are not keeping up with this demand.

In fact, the underlying economic model that built the Internet is breaking. Data from the service providers - those big, global telecommunications carriers that have been building the Internet since the early 1990s - shows that revenue from Internet services has been growing roughly in parallel with the investments required to keep building the Net, allowing service providers to put profits back into Internet extensions.

So far, so good. But the costs to continue building the Net will continue to increase by 18% to 20% every year from now till 2020, while revenue from Internet services will only increase 5% a year in that same timeframe.

Flatter revenue means fewer profits to reinvest in innovation. But, without ongoing innovation, the Internet stagnates -meaning access costs go up, service quality goes down, and growth dies.

We estimate this 'breaking point' may occur as soon as 2014.

Should enterprises care?

Enterprise networks need ongoing innovations in performance, reliability and security just as much as anyone, if not more so.

According to analyst firm Gartner, datacentre consolidation will continue with adoption of more server virtualisation technology and application services, and these will make the datacentre more congested than ever.

Enterprises will also focus on increasing employee productivity with technologies such as unified communication, videoconferencing, etc. And end-users are now demanding more connectivity.

At the same time, enterprise CIOs (chief information officers) are pressured to control operational costs, yet these ambitious goals mean a lot of high-performance networking - in the datacentre, in the campus, and in the branch.

Depending on where you sit in an enterprise, the "Connected Culture" - of device proliferation, content consumption, connected socialisation and machine-to-machine communications - brings a social-media-friendly culture where all things, people, devices, machines, institutions and human knowledge are at our fingertips.

This brings great promise and great peril because it means dealing with demands from an ever increasing variety of users, devices and applications, while increasing performance, security, and reliability.

When it comes to the Internet and our global networks, the Industry "gets it." Industry players know very well that if we don't keep investing in innovation, we're out of business, which is why we know that new approaches are needed for tomorrow's networks.

The typical network architecture found in most datacentres is actually two different networks: A storage area network and an Ethernet backbone.

In both cases there are multiple tiers of autonomous devices that are attempting to co-operate to create "a network." There are also various appliances sprinkled in - applications or security tools trying to solve specific problems.

So, in the end, it is a very complex and expensive environment. This matters, because datacentres drive enterprise networks on the ground, and increasingly, in the cloud.

This kind of complexity is the enemy of performance, security, reliability, and scalability.

Towards a simplified approach

The strategic approach we and the industry have to take, then, is to simplify things and collapse datacentre architectures to enable the exponential scaling in performance, security and reliability needed to meet future networking demands and leverage cloud solutions.

The answer is to virtualise and consolidate security appliances into a single pool of security services then virtualize the access layer - the switches, eliminating the aggregation layer. Then we provide a consistent set of edge services to connect the datacentre to other datacentres, and to the Internet.

The goal is a simple flat datacentre fabric that connects all the elements of the datacentre, yet maintains the simplicity of a single switch, with a single control plane which can scale for cloud-ready applications.

You can then partition this fabric into logical segments, which allows you to virtualise your networks ­- this is essential to achieve exponential gains in performance and scalability - delivering what we call cloud-enabled security.

And because this simplified datacentre fabric operates as a single switch, and is standards-based, it's interoperable for multi-vendor environments.

This is where Juniper sees the datacentre going because there's no other way to scale to meet the network demands enterprises face and truly leverage the cloud.

(Wan Ahmad Kamal is the managing director of Juniper Networks Malaysia, an information technology and computer networking products company)


source: http://techcentral.my/columns/industryviews/story.aspx?file=/2010/7/14/it_col_industryviews/20100714102633&sec=IT_Columns_IndustryViews