This content is part of the Essential Guide: Essential Guide: Application-aware networking
Evaluate Weigh the pros and cons of technologies, products and projects you are considering.

Application delivery optimization: 5 tips for success

Application delivery optimization solutions should make efficient use of bandwidth and increase app performance, but such benefits are missed without these five tips for success.

These Application delivery optimization (ADO) five tips for success show IT professionals how to reap the WAN optimization benefits without sacrificing your career or degrading performance on your enterprise WAN.

The issue: Sometimes application delivery needs help

Users expect optimal WAN performance wherever they are, no matter who they are working with and  whatever platform they are using. Unfortunately, some applications are written in such a way that they perform badly over longer, lossier network links. Transaction flows link one organization's data center with many others, and massive amounts of data need to be replicated among data centers and even between data centers and the cloud. This highlights the point that some applications don’t deal with distance well or are written in ways that don’t make efficient use of bandwidth.

Application delivery optimization (ADO) seeks to fix delivery problems without trying to fix the applications or force end users to limit their options. ADO helps the network become more intelligent and more active in ensuring and optimizing application performance, making sure that everyone gets the performance they need. ADO can help IT meet or exceed user expectations and the needs of the business.

The two main centers of gravity for ADO are WAN optimization—including acceleration, compression, caching, deduplication, traffic shaping and traffic conditioning—and application delivery control, such as load balancing and server offload.  Other ADO solutions include route optimization and WAN aggregation.

Five tips for successful application delivery optimization deployments 

1.     Make sure you understand the problem you are trying to solve before considering application delivery optimization.

User complaints about performance problems are often admirably vague. Just as IT used to have to deal with trouble tickets saying “the Internet is down,” IT often deals now with tickets that say “application Z is slow.”

It is critical that IT staff watch for patterns in the complaints: Is it all apps, some apps or one app? Is it all functions, some functions or one function? With this information, you can triangulate on the actual problem. It may be that the protocol a client uses to communicate with a server performs poorly over the WAN, and WAN optimization will help; it may be that servers are sometimes overloaded handling many encrypted client sessions, and an application delivery controller (ADC) with encryption offload would help; or it may be that a specific function is slow on the server side, and no kind of remediation in the network will fix it.

If you don’t understand the problem, you can buy the wrong solution and find that it doesn’t fix it.

2.     Look at multiple options for solving performance issues.

Look at in-house, appliance-based implementations. Look at virtual application delivery optimization, too. Look at managed solutions that shift the burden of running the appliances to someone else. Look at cloud-based solutions that eliminate the appliances. And whatever delivery method you opt for, be sure to evaluate solutions from four or more vendors, and—if possible—test solutions from at least three of them.

3.     Evaluate application delivery optimization (ADO) solutions with an eye on all your plans.

ADO sits at the confluence of network, security, desktop and data center planning, and it is important that decisions made about optimization not ignore or hinder plans in other areas.  For example, if the company intends to shift to data center-hosted virtual desktop infrastructure (VDI) within three years, should it invest in accelerating delivery of desktop applications to branches? And if the answer is "yes, we can’t wait," then the short timeline should make cloud or managed offerings more attractive. Similarly, if servers are struggling to manage the encryption and decryption of client traffic, the ADO solution would be to offload crypto work to an application delivery controller (ADC). The security architecture might forbid this, or impose additional requirements on any solution picked; the architecture may need to be updated to reflect this new reality. The WAN architecture may call for a shift to Internet VPN connectivity for many branches, which can change the kinds of optimizers that make sense there.

4.     Work across IT silos.

An excellent way to make sure that not just plans and architectures are accounted for but also ongoing operations is to make the team selecting and implementing the ADO solution cross-functional. If staff from security, network, data center and desktop are working together on scoping the need for ADO and identifying a solution, it is far less likely that the chosen method and solution will fail to meet the company’s needs in the end.

5.     Baseline and compare application delivery optimization solutions.

The necessary corollary of understanding the problem and trying multiple solutions is to have some basis for evaluating performance. So, measure throughputs in the affected branches, measure application performance and measure loss and latency.

Interestingly, ADO solutions can usually be put in “watch only” modes that allow you to use them as data gathering engines to establish such performance benchmarks, against which you can measure the effects of compression, acceleration and other services. Cloud services can be offered in a “try before you buy” mode. Having this baseline lets you compare before and after performance, and also compare the differences among the solutions you evaluate.

Continuing to evaluate performance after a solution is in place lets you identify degradation in the benefits provided, which can easily happen as the traffic being optimized shifts over time.

This was last published in August 2011

Dig Deeper on WAN optimization and performance

Start the conversation

Send me notifications when other members comment.

Please create a username to comment.