 |
|
|
Living with a teenager means that there is always someone around to tell you (the parent) that you don't have a clue, but that if you would only listen to them, they would provide you with all the answers. They are younger, they are brasher, but are they ready for full adult responsibilities like they say they are?
The struggle between proponents of SS7 and IP protocols for the telephony signaling market is similar. SS7 is older and more conservative. It goes to work every day, and brings home the bacon every night, all without fuss and without expecting much recognition. IP on the other hand is generating way more excitement for what is really only a part-time job in telephony but, boy, when it gets out there full-time, it is going to show the old geezers how to do it properly!
Only a tiny fraction of the data that is transmitted over networks is signaling, yet it is by far the most precious. Without it, the rest of the data on the network would not know where it was going and how it was going to get there. No matter what the protocol being used, the network itself has to exchange messages to find out information about status and routing. Because signaling is generated internally to the network in most cases, and is therefore invisible to end-users, it “don’t get no respect” (just like fathers of teenagers).
The telephony industry is well into a transition from tone-based, in-band, signaling (e.g. MF tones) to electronic signaling based on SS7. Whereas MF signaling transmits at less than 0.1 kbps and cannot be used during a call, SS7 signaling operates at 56 kbps (over 500 times faster) and can be used during a call. The likely transition to IP will allow operation at much higher speeds and will eventually allow greater integration of telephony and data services. It might not seem that high speeds are required for signaling. After all, if 0.1 kbps was good enough 10 years ago, why are higher speeds needed today and even higher tomorrow? The answer is that services are becoming much more demanding of bandwidth. Short message service is often handled as signaling data, and obviously consists of more data than older signaling systems could ever have handled. In fact, it is even a stretch for SS7 signaling to handle SMS and their day job with the existing network bandwidth!
There are several major reasons why a transition from SS7 signaling to IP is desirable: cost, bandwidth and network integration. IP equipment is generally cheaper because it is produced in much higher volumes. IP is designed to operate at virtually any bandwidth, whereas SS7 was designed around 56/64 kbps data links and has not yet even transitioned to the planned 1.5 Mbps links. Perhaps the biggest advantage will come in the integration of computer communications and telephony. Once integrated signaling exists, applications that provide useful services mixing voice, data and video together will become possible.
However, before IP can take over, it has a lot of growing up to do. SS7 provides unparalleled levels of robustness, a work ethic that IP is still developing. Many SS7 network elements, most notably the STPs (Signaling Transfer Points, equivalent to IP routers) operate in a fully redundant load-sharing mode. When one side of a mated pair fails, the other side takes over right away. Performance might suffer slightly, but the network keeps on ticking, with its users not even noticing.
Another area where IP has some significant problems is in the area of security. Even with encryption, denial of service attacks are a significant problem. This will mean that IP telephony signaling networks will have to be run as virtual private networks, with only limited communications with the public internet, and then only through tightly controlled gateways. For example, PC-to-phone calls might be controlled through the signaling network, but the PC must connect to a gateway that ensures that only legitimate, phone-call-related traffic from the PC gets through. Even then, a rogue PC could take out a gateway through a denial of service attack, eliminating an access point to the network.
Quality of Service is another area where IP still cannot offer the same capabilities as SS7. In the signaling world, QoS is largely based on the probability that a signaling packet will arrive, and the time that it is likely to take to arrive. Unlike many internet applications, a signaling packet that arrives a few seconds later than its expected time may as well not have bothered trying. SS7 cannot offer vast amounts of bandwidth, but it can guarantee that a signaling packet will almost certainly arrive within a small and tight time window. In an IP network, on the other hand, if a router is busy handling other types of traffic, all traffic may get delayed to the point where signaling time-outs occur. Although the IETF is working hard on standards such as MPLS (Multi-Protocol Label Switching), Diffserv and RSVP (Resource Reservation Protocol), there are still no perfect solutions.
The first step in growing up may be accepting a part time job, and IP is heading in this direction, providing physical transport for telephony signaling, but leaving the intelligence at the SS7 and application layers. The IETF protocols M3UA and M2UA support this, allowing signaling messages to be transported over an IP network, but still requiring SS7-aware nodes to initiate, receive and route messages. The STP still performs the primary packet routing functions in this network, although IP routers may be used to route to and from STPs.
A much bigger challenge is to take over the full routing and addressing capabilities of SS7. Conceptually, at least, the domain name server model for determining an address from a telephone number makes more sense than the current SS7 global title methodology. It eliminates restrictions with international routing that currently plague SS7 applications for wireless, and the need to manage data at every router (STP) for each block of addresses for each global title type.
However, domain name servers are oriented to names with a visible structure (i.e www.intertec.com is divided into three parts), but telephone numbers have an invisible structure. North Americans all understand that 14035551212 is really 1 (country code), 403 (area code), 555 (office code), 1212 (line number), but people from other countries may have more trouble isolating the pieces. This is important, because the telephony equipment that ‘owns’ this number for various purposes must be identified by examining a specific number of digits. For example, the first 6 digits of North American wireless phone numbers (7 digits if the country code is included) must generally be examined to determine what HLR contains the mobile’s profile, what Message Center is responsible for routing short messages and what Home MSC should receive calls for the mobile. Other telephony identifiers are similarly variable in structure. The domain name model cannot easily handle such identifiers.
A similar challenge is to allow transparent routing to redundant network elements, such as a fault-tolerant HLR. Should all the network elements sending messages to an HLR know that it consists of 2 or 3 physically separate systems, each with their own IP address? Or, should a virtual IP address be assigned to the HLR, with routers adjacent to the HLR routing based on the current activity status of each its components?
There is no question that SS7 will eventually head for retirement. But, will this be an early retirement, pushed out by the brasher and smarter newcomer – IP? Or, will the industry still be tussling over the best ways to provide security and robustness on IP networks 10 years from now? Certainly the penetration of IP-based telephony systems will be slow until carriers can be certain that a massive failure of their network will not be making news headlines blamed on a rushed and premature transition to a new signaling protocol!
© – Copyright