Sorry about that. I am speaking from a technical perspective.
I was a bit vague here.
Below are some boring specs. Dont bother with understanding what they mean, I just list them for reference. (All available on the WEB) I explain below.
3Com's core builder 9000
Switching Peformance
Initial ATM fabric capacity: 15 Gbps
Initial Gigabit Ethernet fabric capacity: 24 Gbps
Maximum switch fabric capacity (planned): 70 Gbps
Layer 2 forwarding performance: > 100 Mpps
Initial Layer 3 forwarding performance: 25 Mpps
Maximum Layer 3 forwarding performance (planned): 56 Mpps
Cisco's GSR 12000 Family
Scalable Performance
Cisco 12000 Family Cisco 12004 Cisco 12008 Cisco 12012
Bandwidth 5 Gbps 10-40 Gbps 15 to 60 Gbps
Configurable Chassis Slots 4 8 12
OC-3/STM-1 ports1 12 28 44
OC-12/STM-4 ports1. 3 7 11
Cabletrons MSR 8000
Features
Completely standards based
Wire-speed IP/IPX routing
Zero degradation for full functionality
16 Gbps non-blocking switching fabric
15 Mpps routing throughput
Up to 56 10/100 Mbps ports
Up to 14 Gigabit Ethernet ports
2,000,000 Layer-4 application flows
250,000 routes
400,000 MAC Addresses
20,000 security/access filters
Wire-speed Layer-4 application flow switching
Wire-speed application level QoS
Wire-speed Layer 2, 3 and 4 Security Filters
Wire-speed full RMON/RMON2 per port
Wire-speed multicast routing
Redundancy
Hot-swappable interface modules
Comprehensive Java-based management software
OK. What I meant by..
<The 15 or so Yago boxes today are limited in comparison to 3Com or Cisco products>
..was that the MSR box does not currently support anything other than Ethernet, while the competitors will initially have ATM, FDDI and SONET in addition to Ethernet. This does not mean that Cabletron does not plan to have these options, just that they dont currently.
On the other hand, Cabeltron plans to deliver what they have under the veil just a couple of months BEFORE Cisco does. So really, everyones gun is loaded, but its a race to pull the trigger. I was speculating that initially, some bullets may have more bang than others.
Cisco's mid range solution is merely an alteration of the Cat 5k series which is already a rich layer 2 switch. Simply upgrade a board in your layer 2 switch, and you now have a multi-protocol layer 3 switch router.
<Can Cisco deliver that today?>
Actually, the only thing I know of that is being delivered *today* as a working switch router solution is Xylans edge-routing product capable of a feeble 250,000pps of IP forwarding (about 2 100BT connections worth of throughput).
All in all these products are infants if they even exist, so it remains to be seen who can deliver what, and when they can deliver it.
<Comment on Secure Fast and AEOL vs LANE or MOPA over ATM?>
Secure Fast (SFV) is good for small simple workgroups as are most VLAN technologies, but definately not as an end-to-end alternative to routing, as it was originally being sold. The fact that they tried to sell this as a *complete* solution to customers unsuccessfully, seems to have lowered customers image of Cabletron. I was not personally a victim, but they did try very hard to get me to throw out my routers and rely strictly on SFV. Their tactics were discomforting and at times even insulting. But as I stated in an earlier posting, the tactics have recently changed for the better throughout CS.
I cant be specific due to non-disclosure agreements, but I know of 2 large companies which experienced unsuccessful attempts to use Secure Fast VLANS as an alternative to routing. Cabletron was not the only one that couldn't do it mind you... thats why we now have switch routers.
(Kind of reminds me of the days when bridging was supposed to end the need for routing, but everyone quickly learned that routing was still very important.)
Afraid I can't comment much on ATM specific to CS, but I have heard limitations exist in Cabletrons ATM arena (A large amount of which I've read in their own release notes). But I do not have first hand experience with it.
Before Gigabit Ethernet, ATM seemed to be the most viable high-speed backbone solution for the LAN/MAN. But from what I have been reading, it sounds like a lot of people are holding out for Gigabit Ethernet for reasons such as price, familiarity, and comprable services such as layer 4 flow control (application control) and QoS, which is becoming available for Ethernet networks. ATM is still the best solution where definitive control is required for serious integration of voice/video/data, but Gigabit Ethernet will satisfy the requirements of many 'would be' ATM implementations. And of course, ATM fits very nicely in the WAN which is growing fast. |
