WAN Optimisation and Acceleration | Ask The Experts
The Issue
Your network is delivering many business critical applications. You have already deployed optimisation or QoS so that each user gets his-or-her share of the WAN resources.
Has this solved all your issues? Many times the answer is no and this is where Veridical can help you. Since some users are located in far reaching remote sites they are not happy with the response times they are getting from their applications. The users of bandwidth-intensive applications are generally not satisfied with what they receive from their WAN. The WAN delay and bandwidth capacity imposes limitations. Many of our competitors are proposing “box” acceleration solutions to deal with the performance concerns of some of your applications or sites.
Such an approach will rapidly be very difficult to manage and reconcile with global management of the performance of your critical applications over the WAN.
Firstly reassessment of your optimisation and QoS parameters needs to be undertaken. Without a proper allocation of WAN resources, applications will perform badly. Network congestion creates both a lack of bandwidth and extra router delay that jointly contribute to degrading the performance of applications. You cannot count solely on compression to deal with the lack of bandwidth. Compression creates virtual “free” bandwidth, but the amount of bandwidth created varies over time depending on the amount of redundancy in the flows.
Compression is NOT a replacement for advanced QoS methods
Furthermore, deploying acceleration without having implemented proper optimisation can be dangerous for application performance. Acceleration solutions without advanced QoS capabilities can actually worsen congestion when it occurs, especially in MPLS networks with meshed topologies.
- How will your network behave with the new application running ?
- How will the new application affect the other network application’s performance which are also critical to the business ?
- You think that you will need 3 to 6 months and perhaps some consulting help to roll-out this new application correctly but you need some help to justify this spend both in terms of the political situation and also the financial one.
You have already tested the new application in your lab and collected network performance and capacity data to ensure that the network will support the new traffic. To be safe, you also ran a pilot with a limited number of end-users. you will then design adequate policies that guarantee the new application performance while protecting existing ones.
Finally, you have implemented these policies device by device and been monitoring and manually adjusting them. If you have been careful you most probably have over-provision your links and their capacity as its better to over specify than under. The obvious question now is do you know if this new application will seamlessly roll in as per your expectations or do you need something concrete to ensure that it will actually work.
Having verified the QoS implementation, be aware that there are only 3 key bottlenecks that can cause application performance degradations:
- The Application Specific Protocol bottleneck:
- The TCP bottleneck:
- The Bandwidth bottleneck:
Some applications are not designed to run on the WAN since their application protocol is relying on a too high ratio of protocol exchanges over transmitted data. When the network delay is increasing, their performance degrades a lot.
Because of its inherent design, TCP is often not able to utilise all available network resource.
the performance of many (but not all) applications is related to the available bandwidth. If it is higher, then the response time of the application will be lower.
It should be pointed out that not all of those bottlenecks are of equal in terms of their lifeycle. All application protocol bottlenecks can simply be removed simply by redesigning the application to take into account WAN-style network delays. Microsoft is an examply of a company which did this. The newer versions of Exchange combined with Outlook cache mode is for example exhibiting vast performance improvements over its predecessors when used over the WAN. The CIFS protocol was generally known for it very bad WAN performance and has now been upgraded to get rid of legacy WAN performance related issues.
If possible, upgrading the application to benefit from the highest level of WAN compatibility shall always be part of application response time reduction initiatives.
The Solution
However Veridical’s solution is addressing those 3 key application performance bottlenecks through their unique methods.
Our methods of acceleration relieve the WAN performance bottleneck of many applications, including Windows file sharing (CIFS), email (including MS Exchange and Lotus Notes), ERP/CRM applications such as SAP or Oracle/Siebel, database applications and any Web-based applications and more generally any bandwidth-hungry applications including in-house applications and applications based on large file transfers such as CAD applications.
The Veridical solution has ensured the response time of business applications will be minimised while still guaranteeing that the appropriate WAN resources are delivered to each user of critical applications whether they are accessing the application locally or globally.