There are many options available for companies to help manage the at times drastic difference in speed and bandwidth between wide area networks (WAN) and local area networks (LAN). The best way will, however, depend on the applications used and how they behave when transported between the company's offices, and from the offices to the data centres. And as IT departments embrace the cloud, and applications are delivered from both internal resources and third party facilities to users both fixed and mobile, the situation becomes even more complex.
To address some of these issues, there are several application optimisation solutions available in the market to minimise latency, including those by Riverbed and Ipanema Technologies, which accelerate and compress traffic between the company and the data centre, thereby improving the user experience and reducing the need for WAN bandwidth. Over the past few years, this has made a very positive difference to the experience users have of application performance, particularly over high latency (such as satellite) or Internet links.
When it comes to cloud computing applications such as those by Salesforce, Office 365 and Google Apps, the critical factor in determining the user experience is the latency and capacity available over the Internet between the customer and these third parties' data centres. This adds complexity since it is often not clear where the cloud providers are serving their applications from, and the Internet is a mix of many different networks, each of which have different performance parameters.
As always with the cloud, the key consideration for companies is how much control they're willing to relinquish, i.e. what data and traffic they're happy for the SaaS provider to handle in the third party data centre, and what they want to keep on their own turf. We're seeing many companies take care of a subset of cloud-based applications themselves, and hand the rest over to the SaaS provider. Therefore, in these instances, two links are needed - one to the company's private network one to and the public Internet and thence to the SaaS provider.
As mentioned, while we've seen great progress in application acceleration, there's not yet a uniform and efficient way to optimise public Internet traffic because any optimisation technology needs to be installed at both end points - at the company and with the SaaS provider. This creates a difficulty for companies wanting to be able to guarantee a level of application performance to their internal users. To solve this issue, some service providers are offering private VPN connections into key SaaS providers such as Amazon Web Service instead of just using the Internet.
Another challenge that remains - which impacts mainly on large organisations such as financial services businesses that run their own applications - is how companies can optimise their networks so that they have both point to point and any-to-any connectivity between their offices and their data centre, depending on the application requirement. MPLS is great for any to any environments but it is hard to provide deterministic routing through the network for latency- or route-sensitive traffic.
Sign up for Computerworld eNewsletters.