Editor’s Note: Welcome to our weekly Reader Forum section. In an attempt to broaden our interaction with our readers we have created this forum for those with something meaningful to say to the wireless industry. We want to keep this as open as possible, but we maintain some editorial control to keep it free of commercials or attacks. Please send along submissions for this section to our editors at: dmeyer@rcrwireless.com.
Cloud computing is becoming increasingly attractive for mobile operators. However, unlike pure IT, not all types of network functionalities are suitable for a cloud architecture. Mobile data optimization is one example. Recently announced by several optimization vendors, cloud-based solutions for optimization are currently under trial with several operators in the United States with an implementation expected the first quarter of 2013.
Optimization has many different functions that need special consideration in order to reap the greatest benefits from both data savings and quality of experience. These functions include TCP optimization, image compression, buffer tuning, HTTP optimization, Web caching and others.
The good news is that taking data optimization to the cloud does not have to be an all-or-nothing proposition. The decision on how to implement optimization should be based on trade-offs between efficient use of network resources, which is an inherent benefit of cloud architecture, comprehensive reporting and user quality of experience.
When considering implementing optimization for 3G and 4G networks, it is worth analyzing all the options and then striking a balance between risks and rewards before deciding what and how to migrate to the cloud.
Cloud-based caching
Optimizing media once, storing on the cloud and then sharing between sites provides added efficiency for large operators or groups of operators. Using this method, subscribers in the operator’s network can automatically access the most popular content compressed, stored and ready to view, saving network bandwidth and time.
The level of added efficiency, measured by hit rate, is influenced by the number of sites providing optimized content to the cache and the amount of popular content that is in high demand. Generally speaking the hit rate for a central cache comprised from multiple locations can vary from 50% to 70%.
In some cases, where operators experience bottlenecks when delivering video from the cloud, two-tier caching may be the best solution. In this case, some content is stored in the network, and other content is stored in the cloud, as in a mobile-CDN fashion to ensure a quality user experience.
Transcoding
Offline transcoding in the cloud also has great benefits, assuming it is intelligently integrated with the cache. Because the cloud (usually a private cloud) serves many operator sites, it creates a physical cache that represents one huge site with much a higher hit rate and enables more data savings and quality of experience.
In addition, online transcoding – unlike offline transcoding – requires a lot of processing power. As such, it can also benefit greatly from cloud architecture, which can virtualize this processing power and manage it efficiently (for example, when one operator has subscribers in multiple time zones).
Transparency
Although pure cloud configurations may process optimized content more efficiently, they can also threaten transparency, which is needed for full and accurate network reporting.
In a transparent system, if the user browses to a video site and the video site is counting hits for advertising revenues, it will count the hits accurately since it is completely transparent on the user level. Pure cloud system may access the video site as “cloud” and not as the original user, thus limiting the video sites capability to count hits.
How traffic is managed on the client side can determine the level of transparency and amount of data that is available for reporting. Pure cloud optimization systems use methods such as 302 redirect or client software. In the first case, every request is redirected to the cloud, and in the latter case, for example, there is a piece of software running on the mobile handset which sends the traffic to the cloud.
In both these cases, the publisher, for example YouTube, will see the requestor of the content as being the cloud and the user will see the cloud as the source of the content. This masking of the requester and the source of the content can hide statistics that are needed for targeted promotions and network management. In addition it can create a lot of interoperability issues with media applications, which unlike browsers, may or may not support redirections.
With two-tier cloud optimization, traffic is sent to the cloud and optimized, while keeping full transparency for users and content owners – both see the internet as is. In addition, no client software needs to be installed on the subscriber handset. When and how to use the redirect method can be determined by the operator, in order to provide the optimal amount of flexibility.
Full traffic management and advanced optimization
Pure cloud solutions have definite disadvantages when compared with network or inline solutions. Network solutions can optimize all types of TCP traffic, including objects that are sensitive to latency. A good example of this are the dozens of small Web objects presented on each Web page, which compose most of the browsing traffic – sending them to the cloud can degrade speed for the user by 1 to 2 seconds for Web page download and video, sometimes even more. This is exactly contradictory to what operators are trying to achieve with optimization.
In addition, some very valuable optimization methods, such as TCP optimization, must be implemented inline the network since they lose all their RAN-related benefits when in the cloud. In areas where there are large amounts of TCP traffic, it may be advisable to use a network solution for this type of traffic and enable only video and other rich media data to be handled by the cloud.
Cloud configurations significantly reduce the amount of equipment needed for optimization and offload processing power from the operator’s switching centers. However, the best cloud solution is the one that is flexible in a way that enables the operator to decide what processing to do in the switching center and what to do in the cloud to achieve the best fit for superior quality of experience, Internet transparency and data savings.
