Operators blind to true cost of data apps
What are the most inefficient mobile data applications, in terms of an operator’s signaling and radio resources? If you are thinking P2P, mobile video and other http-based applications, then you are wrong.
The most inefficient mobile data applications are mobile email, location based services, secure applications and things like stock updates and tickers.
This is what Alcatel-Lucent’s Mike Schabel, Phd, Alcatel-Lucent General Manager, 9900 Wireless Network Guardian, told journalists in an excellent presentation on the difficulties the huge increases in wireless data usage will give mobile operators.
Schabel said that although attention has been focused on P2P traffic because of the volume involved, such services are in fact very efficient in terms of the network resources they use. In essence, a user downloading a film, or watching a video, gets online, establishes a radio connection, does what they have to, and then gets offline.
This kind of behaviour is predictable and manageable, Schabel said. In contrast, Schabel’s team found that in one operator, mobile email was using 25% of the available signaling capacity, even though it was only responsible for 4% of the network traffic volume. This is because of the constantly on-off nature of a push mobile email device as it receives messages, continuously signing on and off the network. Location based services, that also required a constant “conversation” with the network, are also very resource intensive.
Schabel said that examples such as this show that there is a “hidden cost” in many of the data services that mobile operators are looking to deploy. Operators need to factor in cost per minute and cost per bit, he said. A further cost comes when as a result of not understanding network activity such as this, or the way an application works imperfectly on a device, can cause outages and delays in the network – causing poor user experience.
The answer, he said, is first to understand the root cause of problems occurring within the network, or on other elements upon which the service relies (content/ app server, handset, web server, etc).
Second, operators then can plan and design their networks to meet predictable and known demand (“System engineering 1.0”, in Schabel’s words). This may include additional bearers and resources, but it may also be something as simple as re-aligning existing resources. Only then would operators need to consider other throttling or management methods such as policy management, or CRM and billing tools.
Schabel was speaking to publicise his company’s tool, the 9900 Wireless Network Guardian. This is a device that takes in core network data information in real time, analyses it, and produces data on specific issues of network performance.
Schabel said that the issue to date is that wireless network monitoring tools have been “blind” to IP traffic, while packet inspection and other IP techniques cannot “see” the wireless network. What is needed is a system that marries to two together, so operators can see services operating both in cost per bit, and cost per minute terms.
The 9900 uses algorithms that infer from real time data from the Gn interface what is happening at specific network elements. The technique means that operators do not have to probe all the thousands of network elements on their network to get a view of what is happening, Schabel claimed.
Clearly, test and monitoring vendors will claim a weakness there, but there are also moves to link the IP and wireless sides of the monitoring industry together.
Nokia Siemens Networks is making its second attempt to crack this nut, and is now working with Smarts to provide a single view of wireless and IP networks. Recently, IP test expert Ixia bought Catapult Communications, in a move overtly designed to marry to two sides together.
So Alcatel-Lucent is not alone in attempting to square this circle, although it is taking a different approach.
What was most interesting, though, was the revelation that P2P and other bandwidth-intensive traffic need not be the most “costly” services for mobile operators to support.