|
customer login | partner login |
|
|
 |
 |
||
 |
|||
|
|
 
|
|||
|
|
|
|
Optimizing Portals for Global Use on behalf of end users Portals Magazine, By Brian de Haaff : August 10, 2004 The use of the Internet by organizations to support business-critical Web applications is accelerating. The reason is simple; it is a low cost and ubiquitous technology for connecting with customers, business partners and employees worldwide. Web-based applications such as eCommerce, and online business information services, and portals are helping companies to expand into new markets globally, to increase and protect revenue and to reduce support costs. While the benefits of these applications are clear, the Internet fails to deliver the level of performance and reliability on a consistent basis that is required by the users of these mission-critical applications. In many cases performance is so bad that existing Web portals are unusable, and in many other situations they simply are not rolled out at all despite the clear benefits. New Requirements for Web portals Enterprises looking to leverage portals to streamline business processes face two conflicting design parameters: minimize the cost of application deployments, while ensuring that users around the world have an equivalently satisfactory experience. In short, a company must deploy technologies that transform its use of the Internet into a business-quality information service. The intrinsic value to enterprises of deploying Web portals for domestic and international end users is locked within the technology that mitigates the risks of the Internet while leveraging all of its benefits. The solution must address the following requirements:
Enterprises dedicated to rolling out portals have typically focused on one of two options when attempting to mitigate the substantial risks of users not adopting them due to performance problems. They often consider deployment of the redundant global infrastructure that brings the application closer to the end users or caching static content closer to end users. The management complexity and cost of distributing redundant infrastructure introduces significant costs and risks and is directly in conflict with the need to consolidate. In the recent past, when portals were highly content oriented, the use of Content Delivery Networks (CDNs) or other similar content caching solutions were acceptable options. These technologies were originally implemented to support the rapidly expanding demand for static content and media files over the Internet. CDNs leverage the architecture of the Internet to move a specific class of data as close as possible to the largest possible number of end-users. This class of content data includes files that are frequently accessed but infrequently changed (i.e., company logos, background layouts, large photographs and pre-recorded streaming media). For these reasons, CDNs are not equipped to remove the obstacles to the adoption and management of dynamic Web portals. Neither CDNs nor redundant infrastructure can sufficiently mitigate the risks associated with Web application deployment. The technology that fully addresses these requirements will unlock the value of Web-enabling processes through portals without jeopardizing the fundamental benefits that makes the Internet an attractive medium for process automation. Application Delivery Networks are The Solution ADNs can meet all of these challenges by delivering sub-second response times, increased availability and more efficient asset utilization, and greater visibility into performance worldwide -- with no server or client changes and no capital expenditure. An ADN addresses the needs of enterprises that have applications, content or files that are dynamically generated in a portal or are sporadically accessed and are served from a central location. To better understand the improved end user experience delivered by an ADN, it is instructive to look at actual measurement data from a real support portal for software developers for a large computer manufacturer. End users access the portal that is hosted in Atlanta from various cities around the world. Local users are able to access the site in .63 seconds, users in France in about six seconds and users in Australia in about 12 seconds. This story is typical, the further the end user is from Atlanta, the longer they wait for the page to download or transaction to complete. The root cause of this problem is that the Internet was never designed to support dynamic and highly interactive Web applications. There are three distinct places where delay occurs when end users access Web applications: 1. the "first mile" -- the application processing and Web page response generation time; 2. the "middle mile" -- the transfer time between and within the end user's ISP and the organization's ISP; and, 3. the "last mile" -- the time to send a page of information to an end user via his Internet ISP (e.g., T1, DSL, dial up). Improvements in hardware performance, security, and software design have reduced bottlenecks associated with database access, application processing, and Web page generation in the "first mile". Tier I networks have reduced "middle mile" delays to mere tenths of a second. And due to the prevalence of broadband connections for business users, "last mile" delays are a few tenths of a second. So, why does it take 3-10 seconds (or more) to return a response to an end user? The "middle mile" performance problem for Web applications is caused by distance induced delay which is the interaction between "middle mile" distance based latency, the performance of the Internet protocols, TCP and HTTP(S) and packet loss and congestion within the "middle mile". The problem is not associated with the amount of time it takes for one Internet Protocol packet to travel from one location to another and for a response packet to travel back through the "middle mile" (this is referred to as a round trip time or RTT). While a single round trip across the Internet during normal conditions is usually measured in just milliseconds, the delay is typically multiplied by a factor of 30-50 due to the number of round trips required to transmit an entire Web page or to complete a transaction. The result is that a remote user suffers through many seconds of response time delay during his request. In the event of Internet degradation due to packet loss or congestion, the response time increases by many additional seconds. To address this problem, ADNs maintain bi-nodal network architectures that are used to significantly reduce the number of roundtrips required to download a page or complete a task. This is achieved by having a platform to deploy advanced protocol and optimization technology that improves on the core "best effort" Internet technology. Such a platform also enables the ADN provider to provide visibility into the end user experience from distributed locations around the world and the presence to enable the enterprise to efficiently control its infrastructure use. Every executive or manager responsible for a dynamic Web portal that serves distributed end users should consider an ADN. ADNs have amplified the power of the Internet for portal delivery, effectively unlocking an unprecedented opportunity to successfully Web-enable high value global business processes while simultaneously reducing infrastructure costs. Brian de Haaff is director of product management at Netli, which is dedicated to providing a better Internet for business. He can be contacted at brian@netli.com. |
|
|
|
||
|
||
|
||
|
|
||
|
|
||
|
||