The Great Gigabit Backplane Shootout - Question
#1
The need for higher backplane bandwidth is driven by the increase in aggregate line card bandwidth. This increase is caused by several factors, including an increase in port densities, the integration of new services such as QOS functions, which add more information to the incoming data stream, and faster WAN/LAN ports. While the current market malaise has slowed down the proliferation of high-speed LAN/WAN ports, it has underscored the need for cost efficient solutions with increased aggregate line card bandwidth. This means very high port densities and more service-oriented solutions.
Some customers are asking about a 10Gb/s solution; however, they don't expect to roll it out for several years. The vast majority are looking for 6.25Gb/s transceivers to be integrated in existing or new chassis targeted for next generation products.
Both the need for higher speed systems and the market timeline for such products have been delayed. Higher speed backplanes that can reduce system cost or provide competitive advantage are desired today. However, if the improved throughput is not accompanied by a reduction in cost, the higher speed SERDES products will not be accepted by the market.
For sure, system companies are more cautious in launching new chassis development in a soft economy. Today's BP sweet spot is at 3.125G/2.5G. To achieve meaningful bandwidth increase, every new system will employ state of the art speed, which means 10G.
There are also two practical reasons why 10Gbps backplane is required today:
Clearly the recent market conditions have slowed the rollout of next generation systems supporting increased functionality and/or higher capacity requirements. A greater emphasis has been placed on preserving existing installed chassis based equipment, which in turn has delayed new 10+ Gbit/s SerDes backplanes.
Looking across the system landscape, much of the communications equipment designed in 2000-2001 had just started to make use of 2.5 - 3.125 Gb/s SERDES technology when the current malaise hit. Few SERDES developers have been able to recoup their original R&D investments in this technology since these boxes have not shipped in volume. None the less, both data volume and data rates have continued to climb while IT spending has sagged.
Equipment OEM's such as Cisco face a conundrum. Mike Volpi, Senior Vice President at Cisco, stated at a recent investor's conference that there are over 180,000 Catalyst 6500 systems alone in the field. They are not able to sell new equipment to these customers even as these same customers are requesting to increase the data density of their installed equipment. This means that for the foreseeable future, upgrade solutions that can increase the data handling capabilities of legacy systems will be critical. This is the one segment where IT spending has increased over the last twelve months. Conversations at key equipment manufacturers in this space indicate that this upgrade trend is a strong market reality.
Higher speed transceiver solutions in the next eighteen to twenty-four months that can offer equipment owners "investment protection" will do well even in today's current economic environment. They offer the OEM the possibility of upgrading customer equipment through selective fabric card and linecard upgrades. Customers may upgrade their legacy equipment as needed without incurring the huge cost of total replacement.
Market slow-down has mainly affected the shipments of existing products. Most companies are still investing in the future, recognizing that development of new products is critical for long-term viability. Given this, the design activity using high-speed backplane remains strong.
Absolutely. Today, 3Gbps SerDes can meet the need of most systems with capacity as high as 640Gbps. With the market slowdown, slow deployment of new equipment in the core and the demand for OC-768 interfaces almost disappearing, very few new designs are taking place above 640Gbps. It will take at least 2 more years for 10+ Gbps backplanes to make sense again.
Yes. Clearly the industry slowdown has certainly affected the immediate market demand for higher backplane SERDES technology.
Yes and no. It reduces the ROI, but not the need. The metric used to choose a new system in '04 or '05 will need to improve; size, power, footprint, line card BW, etc. So the systems vendor can either sit on their hands, or innovate. Who will win the RFPs that do come out? Meanwhile the need to leverage existing backplanes is great. An interesting mix.