Subscribe

Is optimisation essential?

Network optimisation has evolved from a bandwidth issue to a focus on application performance.

Stuart Hardy
By Stuart Hardy, business development director of EOH Global Networks Division UK.
Johannesburg, 17 Mar 2016

Optimisation is vital to the functioning of any network. By its very definition, optimisation is the ability to improve the speed and cost in which applications are delivered in a network, or more recently, to and from a global cloud network. Optimisation is the ability to ensure application performance in networks with growing complexity and an increase in the use of bandwidth and applications.

Most of the 'optimisation' technologies that have all but died out over the last few years offered companies a way to reduce expensive bandwidth usage. And, while there is still a component of bandwidth cost savings in most - if not all - optimisation technologies today, it is by no means the driver in customer adoption. Today, that's about application performance.

There are, of course, many technologies all offering up the same opportunity. "Use our services and we will improve your network and application performance, while giving you a powerful insight into the workings of your WAN..."

And, to a large degree, that's what they do. But, unfortunately, no one company has an offering that addresses the broad and evolving needs of customers in all spheres.

So, what should companies be looking for when they consider optimisation solutions?

There are several critical and relevant technologies in optimisation that should be considered when investigating options. The ones that seem more important to a company could assist in identifying the right technology.

TCP IP acceleration

If the company has moved, or is planning to move, its business applications or application development to the global cloud, nothing will have greater importance than dealing with latency, and its severe and detrimental effect on application delivery and user performance.

Simply put, the further away an application is moved from the user, the less available bandwidth there is for that user to access the application. So when an application is moved from within the local data centre to a global cloud platform, like Azure, more than 200ms away, the user experience can and will deteriorate as much as 90%. More bandwidth does not solve this issue.

TCP IP acceleration overcomes this by addressing the inefficiency in the TCP protocol and how it determines TCP window sizing using latency (TCP acceleration). For most companies, especially with globally positioned offices that are struggling with application performance, this one technology can resolve that issue.

Application visibility and control

If a company has a low latency network, such as one based in SA and primarily on last-mile technologies such as fibre and Diginet, etc, then there is a good chance the network problems have less to do with acquiring full optimisation technologies to improve application delivery, and more to do with ensuring the network knows which applications are business-critical and which are not.

WAN optimisation

The term WAN optimisation is generally synonymous with companies and technology that deliver the full optimisation stack. In addition to application visibility and control, and TCP acceleration, in some cases, it includes technology such as compression, caching and de-duplication, all geared to reduce the bandwidth requirements of the network and enhance and improve application delivery. The relevance of these additional services in the wider optimisation category are still relevant in areas like Africa, and in many cases, global networks.

No one company has an offering that addresses the broad and evolving needs of customers in all spheres.

In Africa, the cost of MPLS capacity is still in excess of 15 times that of SA. So, providing a 50% reduction in bandwidth costs is an easy business case when deploying optimisation in comparison to a network and bandwidth upgrade. Global networks are, in many cases, no different. China, Australia, Brazil, Russia as well as many other territories are still extremely expensive to reach using high-quality MPLS networks. In these networks, it's about latency and bandwidth costs to a large degree. But, in South African MPLS networks, where the benefit of TCP acceleration is limited and where the cost of MPLS is falling, it may not be as beneficial to consider the full optimisation stack anymore.

Over the last few years, there has been an explosion of network traffic. With the growing introduction of Internet-based business applications in the WAN (SSL, HTTPS), unified communication, advancements in wireless networking, effects of BYOD, recreational usage, customer wireless zones, virtualisation and cloud, and consumerisation of IT, companies are struggling to ensure application performance to different parts of the business.

Companies can resolve this without having to go down the complex and expensive full optimisation path.

Eighty percent of these challenges can be resolved by providing application-level visibility, classification and control. What's exciting about this space is the coupling with software-defined networking capability and hybrid networking, which allow businesses to use the Internet to supplement expensive MPLS networks. And, with most applications now being accessible from the Internet, companies are starting to re-evaluate their commitment to MPLS networks.

Technology that can assist them to centralise network and application configuration, manage application delivery and leverage the Internet is attractive and available. These optimisation solutions are comparably economic, as they are not weighed down by the expensive compression and acceleration services found in a lot of other optimisation solutions, and in many cases, are software-only services.

Optimisation is an exciting space today, and is growing in relevance. But, finding the right technology can be complicated, as everyone is saying the same thing, even though they all have different capabilities and niches. It's a lot easier when a company understands its problem in terms of some of the discussion points above, before sitting down with the options.

Share