Lars Johan Larsson takes a challenging look at the ways IMS servers can play a vital role in the network infrastructure and how it will be driven forward in future networks.
The fundamental purpose of market analyses is to determine where we are going. In order to do that it is important to know where we have been, where we are and how we got here. I will start with a brief description of some of the prevailing trends in the telecoms industry.
The first and most important one is IP (Internet Protocol). The telecoms industry has decided that IP is the right carrier for all kinds of traffic in the network. After the original analogue communication, there was TDM for digital voice and ATM, which was more data oriented. All of these protocols were designed for their purpose and unique to this environment. Video transfer starts to knock on the door and, at this stage, IP is chosen as the media to carry all the three kinds of information. What is noticeable is that this was the first protocol used that was not designed for the purpose of telecoms applications. IP is a datacom protocol and by using this technology the telecoms industry has selected a component where competence falls outside their own territory.
A bad decision
This was a bad decision. A bad decision should not be confused with a wrong decision. The right decision can be a bad decision if the option is a fatal one. These aspects of good and bad have nothing to do with the merits of the technologies in question. Technologies can be good or bad but this never affects the way the market reacts. The decision was bad because it opened up the whole industry to a new range of competition. IP networks applied for telecoms use have different requirements than in the datacom world, but perhaps not different enough to protect the industry as an isolated entity. I reiterate that the option to this bad decision would have been a fatal one. IP is the right vehicle to deliver the more advanced services that the market will require. Staying with the protocols that were designed for previous generations of networks would have proven far too costly to compete in the new environment.
The second major trend is COTS — Commercial Of The Shelf products. Telecoms servers used to be — and are still — proprietary products that are designed in-house and were differentiators in the competitive environment. This is now changing rapidly. All TEMs have realized that their mission in life is to design and build the best network for their customers, and not to spend exhaustive efforts in building the best boxes to go into the installation. The new standard used is called ATCA (Advanced Telecom Computer Architecture) and most of the TEMs have adopted this strategy. ATCA is specified by PICMG, which has written earlier specifications like Compact PCI and others.
ATCA, as the name indicates, has been developed specifically for the telecoms market and has hit the target at the right time. That is at the time when the potential users are ready to outsource the hardware platforms and focus the resources on areas that have real value to their customers. The transition to COTS has already started, and the first platforms built on the standard are installed in commercial networks. It will take another five years before all telecoms platforms are built on COTS hardware. Some TEMs have a quite new generation of servers on the market and they can not do a shift now, but will wait until the time is right.
The third trend that must be mentioned here is blade servers. This is considered quite a new phenomenon in the IT industry, but it is actually the way telecoms servers have been built for a number of years. The requirements in this market are for scalable and highly available servers and in this respect blade servers have a definite advantage over SMP systems. This discussion was hot a couple of years ago, but the market seems to have lost interest in talking about it. In fact, blade servers are silently increasing its market share and steadily growing in volume. Telecoms is one of the driving markets behind this evolution and my estimation is that this will spread to the rest of the server market over time.
The three trends mentioned should be kept in mind when the conception of IMS is introduced.
IMS is all the rest
Access noes are devices and not servers. These are special designs and include Wireless base stations and DSLAMs. They are produced in high volumes and are optimised designs. In five years time this may very well be a COTS market as well, but the industry is not ready to open this segment yet.
Payload is where all the user traffic resides. Media gateways and GPRS support nodes are best described as embedded computers. This is under the definition that embedded computers are computers that are not computers. The servers must support control functions, sample processing for VoIP and packet processing and inspection. This makes them dedicated to their purpose and, as such, they are the real telecoms servers in this new environment. This will stay a telecoms environment and will be populated by ATCA based servers and probably also dedicated routers for the high speed IP switching.
The next area is IMS, but let’s skip this for the time being and look at the Data Center. This is a pure IT environment and performs duties like billing and customer care functions. The IT environment is, not surprisingly, dominated by the IT vendors and an operator has normally a prime vendor that helps in development and implementation of the IT strategy. This is a controlled climate environment and quite different from the traditional telecoms world.
And finally IMS. By going around it in circles I have simplified the definition of IMS to the point where I can say: IMS is the rest. The 3G partnership program defined three layers above payload that was called Control, Service and Application layers. IMS came in first as a description for multimedia call control, but in practice it was impossible to draw lines like that. As mentioned earlier, a simplification is correct if it serves its purpose. For the purpose of defining the requirement on the servers, this description works.
As opposed to the embedded servers in the payload layer, these are computers. They can be involved in the call control, subscriber management, service delivery or application hosting, but fundamentally they are computers. This is IT technology, but the unique characteristic is that they are placed in a Telecom Environment. Central Office traditionally means operation up to 40°C and built according to NEBS standard. The table is set for a clash between the IT department and the network unit. The first department claims that this is IT technology and they are the masters of this art. The other side argues that this is a business that shall run come heat waves or earthquakes and we know how to deliver services non stop under these conditions.
Opened battlefield
The bottom line is that the result of this battle is by no means given. The TEMs have opened the battlefield by introducing IP in the network and trusted that the operators would still consider this telecoms technology and only trust real hardened equipment in the Central office. As always there is an important factor that comes into play in situations like this. What do the customers say? How do they see the new functions and what strategy will they adopt? So far there is a wait and see attitude although the installations so far have mostly followed the traditional telecoms lines. This is because the operators are not overwhelmed by demands from their customers but pushed more by the telecoms vendors, and they are testing some applications to see if they are the revenue generators they are looking for.
The call control functions are implemented with telecoms technology while service delivery platforms and application servers show a more mixed scenario. This follows traditional lines but, in the long run, there is nothing in the requirements that differs between the various functions. The prime question will be: Is this a business critical application and do I want to pay for guaranteed uptime? The telecoms industry’s implementation of five or six nines availability and the cost associated with that is a key factor. The IT industry also claims that this can be done with commercial IT gear but experience has shown that they are far behind in this respect.
One added factor of uncertainty is that while the areas of IMS and Data Center are well defined in the current situation, this might not be the case in the future. Everything concerned with call control and service delivery will probably stay in the Central Office environment. The scenario with application servers however is not so clear. This is not a traditional element in the network and both IT and telecoms vendors are hedging their bets in this field and offering solutions for both environments.
And the winner is…
Who will be the winner in the end? As mentioned, the customer has the final word and the important factor will be the operators sourcing strategy. The IT vendors want to ‘own’ the data center. The operators will not completely outsource the services and sell the operation to the suppliers. This is a core function and it will stay under the operator’s control. More and more of service and management will be sourced however, in order to focus on the core business.
In the same way, all TEMs have a strategy to sell services. Once upon a time they sold boxes to their customers and the operators had the full responsibility from that point on. Now the strategy is to develop, install, maintain and run a network for the customer. The target here is also to ‘own’ the network and let the operator focus on the revenue generating part of the activity.
The TEMS have opened the battle field and when you are in the business of delivering open technology you must always find a way to close competition out. Ownership in actual or a more virtual ways is one of the best methods. The risk for the TEMs is that if they remain with their own proprietary platforms, such as switches upgraded to IP functionality, and try to offer these in the new environment, the customers will certainly opt for open technology.
The danger for the TEMs is that we will see a PC-fication of the Central Office. The same way the PC has come in and taken over tasks that it was not good enough to perform, commercial (inferior) IT technology will take its place in the IMS arena.