This end of life notice signals the end of ATM. The MGX family of Cisco products are still used by many carriers to provide ATM backbones, but it seems its not enough to sustain the product going forward. And with Cisco refocusing on core business after it’s badly managed forays into toys (Flip Cameras), video creation(Tandberg etc) and Social Media, it really needs to slough off the old skin.
Given that we use Ethernet for nearly everything today including WAN’s, LAN’s, Data Centre, Broadband there is little need to use ATM. On the bright side, this means that carriers will be forced over the next 5 years or so to deploy new backbones that use modern technology instead of attempting to keep the stumbling, complex and low-speed ATM networks running. Yay for customers. Of course there will also be a nice industry of smaller vendors who will continue to supply second-rate ATM equipment to those carriers who stubbornly hold out against investing in new infrastructure – and there are lots of those carriers.
Can we mourn the loss of ATM ?
On the other hand let’s mourn ATM and the opportunities that once often asked for a multi protocol, multiservice, differentiated capabilities and even synchronous/asynchronous services. But because of the cost of when compared to ethernet, we chose to use ethernet everywhere. Having a discussion about relevance of ATM and its superior technology isn’t really worthwhile and we are forced to accept that the use of ethernet, even while accepting its second rate technology, features and capabilities, has ultimately been battered into a shape that works for most people. Even in spite of the IEEE’s best efforts to prevent such things from happening.
In my view, the single biggest mistake occurred was when ATM was standardised — the voice people on the standards committee demanded a cell size of 24 bytes to ensure that voice would get the best possible service. Of course, data was seriously impacted by such a small cell size and even the so-called ‘compromise’ of 48 byte cells meant that voice was better handled than data. Of course it didn’t take long for data to far exceed voice on ATM circuits and ensure ATMs eventual death because of its poor data efficiency.
Another big shout out to all the voice people for being shortsighted. Again.
Sure, ATM was complicated (in its day) and required complex silicon but it has a wide range of features. In the end, we created MPLS to provide circuit like features that ATM had by default, and we created MPLS TE to provide QoS features that ATM had, we munged Ethernet into long haul technology that could be repeated but that ATM had by default, we created 802.1p to emulate the ToS field in ATM and the list goes on and on.
Farewell, ATM. It’s not that you weren’t good enough, it’s just that you weren’t cheap enough compared to Ethernet. Long live the Ethernet king – cheap and good enough™.
It might be worth noting that this is also the end for Frame Relay, since FR and ATM are closely linked in products.