With Long Term Evolution (LTE) services commercially launched in South Africa in October and mass rollout planned before the end of the year, attention has turned to the potential signalling implications of the switch to LTE. In an exclusive interview with HumanIPO, Asaf Weisbrot of F5 Traffix Systems speaks of the challenges of an LTE switch-on, the precautions that should be taken, and the possible costs of failure for operators.
HumanIPO: Please can you explain the implications to a network of switching to LTE services, in terms of signalling traffic?
Asaf Weisbrot: Migrating to LTE networks introduces significant numbers of new network elements communicating solely in Diameter. In parallel, smart phones cause new end user usage patterns that create many more signalling messages. The cross between the two trends generates an estimated 40x more signalling messages for the average network service access. This increase in signalling volume is just the tip of the iceberg and needs to be controlled and managed. Not only are the numbers and signalling peaks higher, but this signalling behaviour is unpredictable.
In addition, moving to LTE requires many signalling peripheral protocols (e.g. RADIUS, web services, LDAP, MAP, and others) to be supported through one signalling controller platform to avoid having signalling islands that do not communicate in the same language until converted to Diameter. All these factors require the use of a sophisticated and flexible Diameter controller that includes a DRA (Diameter Routing Agent), a DEA (Diameter Edge Agent) and IWF (Interworking Function) as early as possible in the LTE migration to allow the operator to contain the broader aspect of signalling in a controlled manner.
What is the meaning of a “signalling storm”?
In general, this means there are suddenly high peaks of signalling messages, multiplied by additional signalling as a side effect (such as time outs, or error messages that explain what happened), which can reach higher levels than the original signalling. Sometimes they are caused by certain network elements going down, and when they are rebooted to go up again, they generate unexpected massive signalling messages. And it’s these inadvertently generated messages can get to such a high level that it’s difficult to find the original message which is the only one of value amongst the others. This situation requires a system that can prioritize the messages to make sure that the original message arrives to its correct destination.
Are such “signalling storms” to be expected when switching to a LTE network?
Signalling storms existed prior to LTE networks and even have been publicized in the press. However, LTE networks are based on Diameter protocol, which is more flexible than legacy protocols such as SS7. The downside of this flexibility is that the protocol and its behaviour is more open to interpretation by vendors and can make the network more prone to signalling storms.
What sort of problems might occur as a result of a signalling storm in principle?
Signalling storms can cause service outages for a specific subscriber, a complete set of services or even bring down a complete network, all outcomes that make bad business for operators.
Is it likely that such a “storm” will occur in South Africa?
South Africa is no different than the rest of world when it comes to LTE networks, smart phones and signalling messages.
What sort of precautions should be taken in advance of switching to a LTE network?
The best precaution that should be taken is the performance of a very careful Diameter signalling solution selection process where the overload protection and failover management are properly tested, including signalling planning that includes testing and simulation prior to launching. Aside from these factors, operators should take into account protecting the network from external and internal security threats, by considering the deployment of a Diameter firewall. The market has just begun to see the beginning of security threats, however LTE is IP-based which opens up the networks to many types of attacks.
If troubles do occur in the launch of a LTE network, how long would it take to repair the network, at what cost?
If a Diameter signalling solution has been chosen carefully, the signalling design process has been thoroughly thought out, and a deep simulation and testing process has been implemented, there is no reason why there should be signalling storm damage, or very minimal at the least. Without thought to a Diameter signalling solution, in the event a full network outage occurs that often leads to customer churn, the costs are extremely high. In our discussions with operators, we’ve heard numbers of $40K – $60K an hour of lost revenues.
