Revving up the WAN

WAN acceleration is becoming a popular subject as IT managers look to reduce the TCO of their infrastructure. But choosing the best solution that provides the level of application performance and ROI desired can be tricky. In a guest column Barry Viles, EVP of operations at IT consultancy, Intergence Systems, provides some tailored guidance.

Tags: InfrastructureIntergence Systems Middle EastUnited Arab EmiratesWAN
  • E-Mail
Revving up the WAN “The potential benefits from deploying WAN acceleration solutions are considerable.” Barry Viles, EVP of operations, Intergence Systems.
By  Barry Viles Published  September 20, 2009 Network Middle East Logo

There are myriad Wide Area Networks (WAN) acceleration products and solutions in the market place today, all looking to address the diverse range of network performance and operational cost challenges facing commercial and governmental organisations. These challenges might include how to significantly improve response times for remote applications, especially over high latency links; LAN-like performance over the WAN for remote access workers, irrespective of location. The hurdles also include how to defer WAN upgrade OPEX costs by enabling more headroom on existing WAN links to accommodate traffic growth and how to maintain end-user application response times within a consolidated and centralised data centre environment in the drive to reduce operational costs and improve efficiency.

Furthermore there is also the challenge of how to go about speeding up the completion of large bulk data transfers and disaster recovery across the WAN, to minimise data loss and achieve shorter recovery and time objectives.

The potential benefits from deploying WAN acceleration solutions are considerable.  For example, currently around 31% of IT budgets are consumed by monthly WAN costs and the introduction of WAN acceleration can contribute WAN savings of between 50% and 90%, within a typical ROI period of seven to 10 months. At the application level, 10mb Power Point presentations can be opened by remote users around 35% faster with the introduction of WAN acceleration, with an 85%-90% average data rate reduction when carried across the WAN.  And there are plenty of other equally compelling statistics like these available within the WAN acceleration industry to encourage adoption.

What's the problem?

Firstly, before going into solution mode, we need to understand what business challenges and problems need to be resolved by the introduction of a WAN acceleration solution and what is the outcome expectation in terms of application performance improvement, user expectation around business productivity increases, the bottom-line ROI and the pay-back period.

At a general level businesses are looking to achieve cost and business performance benefits by adopting end-to-end networked application acceleration solutions to reduce the amount of data sent over the WAN, ensure the WAN link is never idle if there is data to be forwarded across it, prioritise business critical and delay sensitive traffic over all other traffic sharing the link and reduce the number of round-trips or application hand-shake turns across the WAN for a given application transaction. It is important for businesses to optimise the end-to-end path of a transaction, through the WAN, the LAN, the data centre and server environments.

These objectives can best be achieved by first understanding the current application mix flowing across the WAN, the levels of WAN utilisation at peak periods, identification of performance pinch points and the bottle-necks for critical business applications.

This ‘base lining' of current network performance will allow you to focus on your organisation's acceleration needs with respect to business applications, as well as maximising the use of the enterprise infrastructure prior to WAN acceleration.  For example within an enterprise, it is typical to observe up to 50% of network bandwidth being wasted on non-business, recreational or malicious traffic, meaning that it's not uncommon for time critical business transactional applications such as SAP to be fighting for bandwidth with You- Tube and iTunes downloads.  Organisations apply different policies in respect of managing non-business application usage, but most would agree that we should be seeking to control social network traffic levels across the enterprise within agreed or ‘acceptable' levels, certainly not to accelerate them.

What about the future?

We also need to plan for the future when considering the introduction of WAN acceleration. Applications continue to evolve and develop at a rapid pace, and the mix of protocols traversing an enterprise network changing almost monthly. There are two key trends that are emerging, today. The first of these is the ‘Webification' of applications. This refers to the growing movement to implement web-based user interfaces for web-enabled applications that will utilise ‘chatty' web-specific protocols (i.e. requiring tens, if not hundreds, of WAN handshake transmissions for a single information transaction), such as HTTP.  In addition, protocols such as XML are ‘dense' protocols, meaning communications that are based on XML consume more IT resources and processing power than non-XML based solutions.

The other trend that we can see raising its head is the outsourcing of enterprise applications. This refers to organisations relying upon third party hosting providers to run their web-enabled in-house applications, or the use of public domain hosted applications such as, WebEx, Oracle On Demand, CaseCentral or Digital Insight. So any WAN acceleration solution under consideration should be able to accelerate web applications, and increasingly, both internal and outsourced SSL-encrypted web applications as well.

Options and solutions

WAN acceleration works where a device - hardware appliance or software - intercepts traffic destined for the WAN, compressing and optimising it so that only the minimum amount of required information has to be transmitted to the remote site, which will also have a WAN accelerator installed.

Solutions can be hardware appliance-based (WOC: WAN Optimisation Controllers), virtual software-based, Soft-WOC, or operate either symmetrically - solution device at the data centre and each branch site - or asymmetrically.

In broad terms, solutions address latency challenges and opportunities to free-up more available WAN bandwidth at three distinct levels. Firstly, data-streamlining, which is to reduce the amount of repetitive and unnecessary traffic being sent across the WAN. This might release up to five times more available bandwidth headroom that can then be utilised to absorb future application traffic growth.  Secondly, there is transport streamlining which helps to accelerate applications by removing common protocol transmission ‘chattiness' and latency constraints such as TCP windowing. And lastly is application streamlining, which again helps accelerate applications by removing specific protocol ‘chattiness' and latency, applicable to applications such as CIFS (file sharing), HTTP, HTTPS and SSL web services, email services (MAPI), and many more. The combination of transport and application streamlining can lead to application performance increases approaching one hundred times.

In order to achieve these application performance improvements end-users can employ a variety of techniques within the solutions, such as caching, compression, data de-duplication and streamlining, protocol optimisation and traffic prioritisation and shaping.

Selecting the right solution

Once a solution has been mapped on paper against business needs, including any return on investment business-case calculations, the next stage is to pilot the WAN acceleration solution on the bench, or preferably on the network, with live user applications. If there appears to be two or more potential competing solutions, then a ‘bake-off' evaluation may be required.

Piloting allows you to verify whether or not the right solution is being selected prior to wholesale deployment, and gives you an opportunity to verify the extent to which anticipated bandwidth savings and application performance improvements look to be achievable, specifically within the context of your own operating environment. Key areas to evaluate include performance improvements per application; for example how efficiently CIFS traffic is being transferred across the WAN. They also include an evaluation of operational features, such as high-availability, traffic transparency (or tunneling), connectivity with regards to in-path, out-of-path, clustering, pass-through failure mode. In addition to that usability - ease and duration of installation, configuration of the device, how easy it is to download a new software image to multiple acceleration devices, or move to another server blade, etc - can be considered along with the manageability of the solution which encompasses the ease of use of the command line interface, reporting, GUI and web interface.

It is all too easy to focus purely on the application performance and WAN bandwidth saving aspects of the solution, and ignore the usability and manageability aspects. In a recent survey across several hundred IT professionals, 73% claimed that in most cases it was the end-user who first noticed a performance degradation of their business applications - not their IT organisation. So ensuring that the WAN acceleration solution remains fully optimised for maximum performance is arguably as important post deployment, as was the solution selection process itself.

Barry Viles spent 10 years at Equant Network Services joining in 1990, and managed the European Pre-Sales operation before becoming Vice President of Service Delivery at Intergence. From 2000 to 2007 he worked at BT Radianz, and as Head of GFS Global Engineering, left to join Intergence Systems.  Barry is responsible for the successful delivery of the network optimisation projects and works closely with intermediary service providers and end-user clients.


Situations that can potentially benefit from WAN acceleration:

* Organisations with central data centres serving remote branch offices
* Remote workers / sites with poor or high latency connections
* Satellite and mobile workers * Sites suffering from congested links
* Applications appearing to ‘freeze' or act jerkily, particularly at peak usage periods
* Organisations wishing to consolidate their infrastructure into a set of virtualised central application servers

Add a Comment

Your display name This field is mandatory

Your e-mail address This field is mandatory (Your e-mail address won't be published)

Security code