Telecom infrastructure development is moving into a third phase of infrastructure development: the shift to pre-integrated systems that include high availability middleware, protocol stacks, unified management, and hardware platforms. These systems offer a quick path to market since the equipment manufacturer develops only the application that runs on the system.
The first phase of infrastructure development had telecom equipment manufacturers (TEMs) using a vertically-integrated approach: designing their own silicon, building custom chassis and boards, and writing all of the software in the system, including operating systems, databases, protocol stacks, and applications.
The second phase of infrastructure development was marked by a shift to the commercial off the shelf (COTS) model where the individual blades and software modules are procured from different vendors and integrated together by the TEM. The COTS model offers an improvement in time-to-market and lower development costs but is not without problems. In particular, the time it takes to integrate the components has become a significant part of the overall schedule, and it has proven difficult to reduce time-to-market below 24 months because of this integration time.
This desire for even quicker time-to-market has given rise to a third model, the pre-integrated system. Pre-integrated systems feature a comprehensive software stack built on a standards-based hardware platform, designed and tuned for particular network elements. In the best examples of this third phase of infrastructure development, the system includes reference applications, protocol stacks, unified management, high availability middleware, platform essential services, and a competitive hardware platform. An example of an integrated system, the Continuous Computing FlexTCA system for IPTV, Security Gateways, and Wireless Core Infrastructure is shown in Figure 1.
Figure 1
Once the decision has been made to use a pre-integrated system, here are some strategies to ensure that the time-to-market advantage is preserved:
Do:
Understand how applications should integrate onto the platform
Every platform makes assumptions about how applications should integrate onto the platform: what APIs are available, how the application should be made highly available, and how the application can be tuned to the platform for maximum performance. These assumptions affect whether the platform is a good fit for the application, and trying to shoe-horn an application that violates the assumptions is a recipe for disaster.
Understand the support model for the various system components.
Two models exist among system vendors:
- One stop shop: the system provider provides all of the front line support and tracks support issues, patches, and fixes internally, providing a single interface to the customer
- Reference sell model: the system provider partners with multiple software vendors for a total solution, but the sales relationship and support for each component is negotiated with the individual vendors. Unless carefully managed, this has the potential to cause finger pointing when complex issues arise
Take advantage of the vendor's ability to provide professional services to accelerate development
Some system vendors offer professional services and are experts in particular applications. Take advantage of this expertise to create sample applications, benchmarks, or integrate existing code on the system platform. All's fair when the name of the game is time-to-market!
Ensure the system is standards-compliant
Vendors are always tempted to create customer lock-in by making aspects of the system proprietary. Insist that the system conform to standards wherever they exist " this ensures that the system can be evolved over multiple generations while preserving the advantages of a pre-integrated system to get the initial products to market
Probe the high availability strategy for the system
High availability is the most complex portion of an integrated system, as availability must be ensured for the application itself, the protocol stacks, system management components, databases, network links, and blades. Understanding how the different elements are synchronized and managed together is vital to ensure the overall application availability targets can be met.
