By some estimates, the amount of information available on the Internet is growing anywhere from three to five times per year. This does not even take into account that the bulk of today's business content sits behind the corporate firewall.
Concurrent with the information explosion, enterprises are seeing changes in the workforce. For example, 30 years ago, it was safe for U.S. companies to assume that English was the primary language of an employee. This is no longer true. In addition, enterprises are using more temporary workers and seeing higher turnover rates for full-time employees. The result is that effective communication is essential to maintaining high levels of productivity.
This change in workforce demographics means effective communication is no longer about more information, but rather better information.
Streaming media offers companies ways of communicating their messages and strategies through multimedia, which has been proven to be a more effective than traditional text- and graphics-based methods. Video-on-demand (VoD) applications can deliver targeted multimedia information that is customizable to be in different languages and more effective for training in complex tasks. For example, regardless of language, showing a technician how to perform a mechanical repair is often more effective than explaining the procedure in a document.
Streaming refers to the way multimedia content is delivered to and "played back" for end users. Before streaming, multimedia files were downloaded in their entirety before viewing. The large multimedia file sizes generally translated into painfully long wait times for content. Streaming technologies allow users to view and/or listen to multimedia content as it is downloaded, reducing the time before playback begins.
An early approach to streaming, called "progressive downloading" simply breaks up media files into chunks that can be played back as they arrive. These chunks are delivered via HTTP, requiring very little change to existing infrastructure. However, because the HTTP protocol was designed for much smaller, static objects, progressive downloading cannot respond to changing network conditions or provide delivery for "live" content. In addition, progressive download only allows for serial playback--there is no way to provide the typical VCR functions most end-users have come to expect. Examples of common progressive download technologies include Macromedia's (www.macromedia.com) Flash and Shockwave, Microsoft Corp.'s (www.microsoft.com) AVI (Audio Video Interleaved) format, and standard MPEG formats such as MP3 files.
Real-time streaming built on the concept of progressive download, provides a solution that allows for a more fully functional multimedia environment. Real-time streaming technology uses specialized protocols and content formats that can adapt to network conditions, respond to VCR-like input from end users, and provide delivery for live and VoD content. The most common streaming media products are RealNetworks Inc.'s (www.realnetworks.com) RealSystem, Microsoft's Windows Media Technologies, and Apple's (www.apple.com) QuickTime. In addition, standards-based file format (MPEG4) and transmission protocol (Real-Time Protocol, or RTP, and Real-Time Streaming Protocol, or RTSP) efforts are underway.
Quality of Experience
In order to take advantage of the communication efficiencies of multimedia, deployments of streaming applications require the same smooth, uninterrupted audio and video available from TV and radio. In other words, the quality of the end user's experience must be reasonable.
The first important piece of a streaming application is the content. Without productive content, no amount of infrastructure optimization can make the application useful. Most enterprises get their content from a variety of sources. Training material can be bought "off-the-shelf" from companies specializing in online course content. For material that is specialized to an organization's business and for live content, a variety of third-party production houses can provide support for scripting, filming, and promotional activity. These vendors range from event-production service providers like Akamai Technologies (www.akamai.com), Yahoo! Broadcast (www.yahoo.com) and Activate (www. activate.com), to smaller, often local, boutique production companies.
From the quality of experience (QoE) standpoint, the most important aspect of production is encoding, or the process of putting content into streaming formats. Generally, multimedia streams are encoded for delivery at a specific bit rate. End-user bandwidth capability typically ranges from modem speeds to DSL connections, to high-capacity LAN links. Sometimes, especially for content with a broad audience, it is impossible to control the connection speed of the user population. As a result, many organizations choose to make multiple separate versions of a given streaming file available. This involves encoding the file at a variety of bit rates and providing links to the different outputs.
While encoding multiple bit rates can allow content providers to address an audience with varying connections, it does not solve the problem of unexpected network congestion. RealNetworks and Microsoft have added functionality to address this issue. RealSystems' SureStream and Microsoft's Intelligent Streaming technologies can reduce the delivered bit rate when network congestion occurs. If network congestion clears, the technologies can then dynamically increase the delivery bit rate.
While streaming media can address QoE from the encoding perspective, the reality is that network congestion often interrupts streaming transmissions, introducing jittery video and audio static, and potentially disrupting performance of existing applications. To prevent this, an infrastructure solution that ensures quality of experience and preserves service for existing applications is required.
For live broadcasts, a networking technique called IP Multicasting, from traditional packet networking vendors like Cisco Systems Inc. (www.cisco.com), can be used to conserve bandwidth. With multicasting, origin servers send streams to an IP multicast group. The underlying packet network (routers and switches) duplicates packets where needed, resulting in lower bandwidth consumption when compared to "unicast" delivery.
A few caveats apply to IP multicasting. First, IP multicast requires special multicast-enabled networking equipment. The effort involved in enabling multicast on a network can range from software updates to hardware replacements. Second, since end users have no control of the data stream, multicasting only supports live content and end-users can only connect to or disconnect from a multicast stream. There is no way to stop, pause, rewind, or advance. Finally, multicasting solutions do not support intelligent encoding schemes. This implies that a multicast streaming application has a limited capability to respond to changing network conditions or support diverse end-user populations.
As a result, a growing number of enterprises are enabling live and VoD streaming applications by deploying a Content Delivery Network (CDN) using solutions from vendors like CacheFlow Inc. (www.cacheflow.com) and Network Appliance (www.netapp.com). CDNs distribute content closer to users, allowing streaming content to be served on LANs that typically are less subject to congestion than corporate WANs or the Internet backbone.
CDNs enable organizations to provide a high QoE in primarily two ways: stream splitting and stream caching.
Stream splitting retrieves a single live broadcast stream from the originating server and "splits" the stream for delivery to all requesting users, thus eliminating the need for multiple, redundant transmissions of the same content. The reduction of in-bound streams from one-to-many to one-to-one saves a significant amount of bandwidth and enables a larger audience to enjoy high-quality access to live content.
Stream caching allows for frequently requested on-demand content to be stored locally, close to users. As a result, content is retrieved from the origin once but can be used to serve multiple requests. When an edge delivery node receives request for on-demand content, it checks to see whether the requested media already resides in its local cache. If the media is not already in the cache, it acquires the media file from the server and simultaneously delivers it to the requesting client.
A final aspect of effective streaming applications is content and usage management. In training applications, learning coordinators need to track progress and anticipate future training requirements. For external events, promoters want to track viewer demographics and gather sales prospecting information. On the consumer side, content providers need to track subscriptions and usage, as well as control access to material that requires advance payment. These requirements imply a need to understand content and users, as well as provide appropriate integration with back-end publishing, billing, learning management, and security systems. For all of these applications, the capability to manage, update and verify content is critical.
Streaming media can allow enterprises to improve the quality of internal communications and extend the reach of external communications, but the networking requirements are potentially disruptive to existing applications. Companies interested in reaping the benefits of streaming applications need a solution that can deliver a high QoE and provide the content management and back-end integration capabilities required to effectively manage their streaming content.
| The CDN Premise | |
| Stream Splitting | Stream Caching |
|
|
Source: CacheFlow Inc. (www.cacheflow.com) | |
Mark Kraynak is strategic marketing manager for CacheFlow Inc.
(www.cacheflow.com).
He can be reached at (408) 220-2200 or
mark.kraynak@cacheflow.com.
