Finally, someone who understands the ATM market...
Thomas C. White writes:
>>> RBOCs and major independent telcos (largest overall switch users) originally conceived plan back maybe 6 years ago to implement ATM networks by having major CO switch providers such as NT, ATT, etc build ATM functionality into the "big iron" CO switches (Class 5, Tandem etc.). This fit with their overall telco network model. So all these high capacity switch mfrs chased endlessly on this business model. FYI, Fujitsu FETEX product was actually the first CO level switch to have native ATM capability (they had developed probably first proprietary ATM chipsets). Fujitsu spent about 4 years and probably $100 million to homologate into RBOCs. This was a dismal failure .>>>
Fujitsu at the time (1992-3) paraded FETEX model 150 "wins" in networks which were really experimental test beds. Everyone, cheered on by the newly formed ATM Forum, was all hot and bothered with the prospect of ATM taking over the CO voice infrastructure world and were busy testing everything with ATM inscribed on it. ATM was the technological fix for everyone's woes.
It just didn't happen, as RBOC's are generally unwilling to experiment on real live (i.e. paying) customers with unproven technologies, especially one which requires a clever network management system to arbitrate traffic flow. ps Even today, very little voice traffic has made the transition to 53 byte cells.
Enter Video on Demand, the 1993-4 equivalent of "what every marketer thinks the common man (and woman) wants." All the RBOC's, most cable companies, and 14 other entities who had no business being in the business thought that VoD was The Next Big Thing. That drove the next phase of ATM development, but VoD crashed and the 'Net became the defacto Next Big Thing. (n.b. without the aid of anyone's market research department but with the assistance of some youngsters at CERN and NCSA.)
The trouble started in 1994-5 when ATM switches designed for VoD were used in data applications, esp. with that troublesome TCP/IP protocol. All that bursty traffic broke buffering sizes and schemes designed for CBR video streams. StataCom, a big FR supplier, was the first to try to fix this problem, fundamentally by adding huge buffers to their ATM interface cards. Note that at this time, AT&T/Lucent had the largest available switch (20Gb/s) and may still, Fore was just playing in the sandbox with a converted VME/SparcEngine chassis, and GDC may have had a reasonable product but lacked large domestic customers. Cascade was not an ATM player, since their raison d'etre was FR (pretty smart since FR was growing faster anyway). Little by little, CSCC was chipping away at STRM's FR market share, and although keeping an eye on ATM it wasn't overly concerned that it had no product as the ATM market was miniscule.
I digress for a moment; about this time (1995-6) the concept of "overlay network" came to be. In situ providers (i.e. RBOC's) would have two networks, one for voice and the other for data. Some newcomers (the MFS's of the world) would have data-only networks which could be used for transport but not necessarily local switching. For both of these situations, FR to ATM aggregation became a neat idea, since ATM switching speeds could be used to route FR data. To wit, two companies virtually owned the FR waterfront: STRM and CSCC. CSCO saw this early and snapped up the cheaper of the two (April 96) which gave them instant CO penetration and access to maybe two thirds of the world's data (LAN, MAN & WAN). CSCC meanwhile, introduced an ATM switch (the 500) and also a RA concentrator for the legions of unwashed who were joining the 'Net every day. Smart moves, especially since RBOC's which had CSCC FR equipment were likely to hang CSCC ATM equipment off of it especially if they could be managed through the same interface.
So now we have two infrastructure networks, voice and data, and the collision thereof has been postponed until about 1999. Which bodes well for CSCO and CSCC as the growth of overlay networks will help both. And that same growth is helping ATM kingpin and FR wannabe Newbridge, remote access king Ascend, and a myriad of other players. Which you already knew. But my experience is as follows. If I run a large data network for real customers and expect good performance with minimal downtime, I am going to use Lucent, StrataCom, Newbridge, and Cascade as my "core" (read: >10Gb) switch providers. Edge devices for FR aggregation and protocol conversion (ADSL, ISDN, and sync T1/T3 to ATM) will be from Ascend, Cisco, Fore, and GDC to some extent. That is the situation today. Invest wisely.
Jim
sort-of-techy-ATM-guy-with-small-positions-in-all-named-above |
