You're talking about usage -- which is commonly oversubscribed -- whereas I was talking manufacturer's specs.
Assuming "oversubscribing" is not an issue here: Speccing for "almost 200Gbps" in this situation, is overkill.
Of course, I know that ISPs stretch this line of thought to the extreme, which leads to pissed of customers. But somewhere in the middle is the way to go. IMO.
OK. They will most likely have to decide which switches to use and purchase them before 2010 if they are to be done by then. But claiming that they will have to pay this and this much based on today's prices is not very likely. I think it is fair to assume that they can get a lot of work done before they actually have to purchase a lot of the equipment which will get drastically cheaper in, say, 3 years.
I basically agree with what you are trying to say, but you mess it up with your hypothesis.
1 Gbps is nice. Now pump an entire apartment unit with GE into the switch. What speed will the internal switch fabric have to support? Assume 200 apartment units, then that is in the neighborhood of 200 Gbps of switch fabric throughput. Consider most of the traffic will be going OUT of the building, the outside pipe will have to be something like an OC-48 ATM or 10-G ethernet connection.
This is not even remotely likely. Analogue to electric power engineering: They will more than likely assume that, say by a factor of perhaps 0.6, not all users will be using the full capacity of their bandwidth at all times. Thus the sum is more like a total of 120Gbps.
And the traffic won't most likely be going OUT of the building. But I assume that what you mean is that most of the traffic won't be between apartments. Lets say only 5% of the traffic will be. Now you're looking at 114Gbps.
Still, this is a lot. And as I said, I basically agree. But the figures.
Now, of course one might argue that if the network in this building is designed around 114Gbps throughput, it will not be true 100Mbps. But these kind of design criterias are very common, and I bet they will be used here. It is simply a matter of cost. The likelyhood of that all the users will need their full bandwidth at all times is so small compared to the cost of implementing it that the whole thing becomes blatantly obvious.
Secondly, this "expensive" equipment won't, as stated in earlier posts, be as pricey 7 years from now.
Presently, you're right. It's far too bulky to be considered for anyone but die-hard geeks. Similar to someone hundreds of years ago considering lugging around a grandfather clock on their wrist. Or thinking of carrying a phone in your pocket thirty years ago. Eventually, it becomes cheap enough and small enough where it makes sense.
Agreed. It is of course good that some pioneers show us how it can be, but I'd much rather have implants.
Integration is all good, but as for integrating functions into clothing I believe it can be at the expense of flexibility.
I would much rather have a lot of functions integrated in my mobile, and be able to bring those functions with me in situations that I might choose another attire.
Nevertheless: this submission had actual links. And this, correct me if I am wrong, was the first time a working prototype has been officially demonstrated.
Slashdot Mirror
You're talking about usage -- which is commonly oversubscribed -- whereas I was talking manufacturer's specs.
Assuming "oversubscribing" is not an issue here: Speccing for "almost 200Gbps" in this situation, is overkill.
Of course, I know that ISPs stretch this line of thought to the extreme, which leads to pissed of customers. But somewhere in the middle is the way to go. IMO.
OK. They will most likely have to decide which switches to use and purchase them before 2010 if they are to be done by then. But claiming that they will have to pay this and this much based on today's prices is not very likely. I think it is fair to assume that they can get a lot of work done before they actually have to purchase a lot of the equipment which will get drastically cheaper in, say, 3 years.
I basically agree with what you are trying to say, but you mess it up with your hypothesis.
1 Gbps is nice. Now pump an entire apartment unit with GE into the switch. What speed will the internal switch fabric have to support? Assume 200 apartment units, then that is in the neighborhood of 200 Gbps of switch fabric throughput. Consider most of the traffic will be going OUT of the building, the outside pipe will have to be something like an OC-48 ATM or 10-G ethernet connection.
This is not even remotely likely. Analogue to electric power engineering: They will more than likely assume that, say by a factor of perhaps 0.6, not all users will be using the full capacity of their bandwidth at all times. Thus the sum is more like a total of 120Gbps.
And the traffic won't most likely be going OUT of the building. But I assume that what you mean is that most of the traffic won't be between apartments. Lets say only 5% of the traffic will be. Now you're looking at 114Gbps.
Still, this is a lot. And as I said, I basically agree. But the figures.
Now, of course one might argue that if the network in this building is designed around 114Gbps throughput, it will not be true 100Mbps. But these kind of design criterias are very common, and I bet they will be used here. It is simply a matter of cost. The likelyhood of that all the users will need their full bandwidth at all times is so small compared to the cost of implementing it that the whole thing becomes blatantly obvious.
Secondly, this "expensive" equipment won't, as stated in earlier posts, be as pricey 7 years from now.
Presently, you're right. It's far too bulky to be considered for anyone but die-hard geeks. Similar to someone hundreds of years ago considering lugging around a grandfather clock on their wrist. Or thinking of carrying a phone in your pocket thirty years ago. Eventually, it becomes cheap enough and small enough where it makes sense. Agreed. It is of course good that some pioneers show us how it can be, but I'd much rather have implants.
Integration is all good, but as for integrating functions into clothing I believe it can be at the expense of flexibility. I would much rather have a lot of functions integrated in my mobile, and be able to bring those functions with me in situations that I might choose another attire.
Nevertheless: this submission had actual links. And this, correct me if I am wrong, was the first time a working prototype has been officially demonstrated.