- 注册
- 2012-02-27
- 消息
- 9,440
- 荣誉分数
- 2,120
- 声望点数
- 273
28M/1M的 Cable internet:
现在用的Cable modem是Docsis 2.0的，有必要升级到支持Docsis 3.0的吗？
还是等等直接买Fiber Optic Modem？
==========================
Keep in mind that the difference between DOCSIS 2.0 and DOCSIS 3.0 was only two things: channel bonding for both the upstream and the downstream, and support for QAM128 for the upstream. Downstream both use QAM256 so the maximum downstream throughput for a single 6MHz channel is 42Mb. QAM64 upstream is 30Mb, and QAM128 should be something like 35Mb (I don't have the specific number in front of me), but I've never seen anyone use QAM128. So for all practical purposes the only difference is channel bonding.
Given that most services are still well below 42/30, the advantage to channel bonding isn't really throughput, it's load balancing. DOCSIS 2.0 only supports a single channel, so 2-3 users on a fast connection can easily saturate a 6MHz channel. Since DOCSIS can't seamlessly reallocate on the fly (DCC requires a modem retrain), you're basically stuck with the people on your channel, so you better hope not everyone is trying to use their connections at once.
With channel bonding on the other hand, you'll be using 3-4 channels. The worst case scenario (everyone on the service trying to max out their connection) is still the same, but with a larger pool of users the average isn't as bad because your connection can't be dragged down by a very small number of the wrong people. Thus performance is more consistent on average. In the long run of course this enables higher performance beyond 42/30. The secondary benefit is that with systems like PowerBoost, bonded channels mean not only can users boost higher, but for the same reason that bonding benefits the average speed, bonding makes it easier to boost big by siphoning a bit of bandwidth off of multiple channels.
It's because we're talking about the average case however that not everyone sees immediate benefits. Most users with PowerBoost available should see some kind of benefit, otherwise the biggest benefit will be for those users with a particularly unlucky channel allocated to them. And of course this assumes a DOCSIS 3.0 headend; none of these benefits would be seen with a DOCSIS 2.0 headend.
现在用的Cable modem是Docsis 2.0的，有必要升级到支持Docsis 3.0的吗？
还是等等直接买Fiber Optic Modem？
==========================
Keep in mind that the difference between DOCSIS 2.0 and DOCSIS 3.0 was only two things: channel bonding for both the upstream and the downstream, and support for QAM128 for the upstream. Downstream both use QAM256 so the maximum downstream throughput for a single 6MHz channel is 42Mb. QAM64 upstream is 30Mb, and QAM128 should be something like 35Mb (I don't have the specific number in front of me), but I've never seen anyone use QAM128. So for all practical purposes the only difference is channel bonding.
Given that most services are still well below 42/30, the advantage to channel bonding isn't really throughput, it's load balancing. DOCSIS 2.0 only supports a single channel, so 2-3 users on a fast connection can easily saturate a 6MHz channel. Since DOCSIS can't seamlessly reallocate on the fly (DCC requires a modem retrain), you're basically stuck with the people on your channel, so you better hope not everyone is trying to use their connections at once.
With channel bonding on the other hand, you'll be using 3-4 channels. The worst case scenario (everyone on the service trying to max out their connection) is still the same, but with a larger pool of users the average isn't as bad because your connection can't be dragged down by a very small number of the wrong people. Thus performance is more consistent on average. In the long run of course this enables higher performance beyond 42/30. The secondary benefit is that with systems like PowerBoost, bonded channels mean not only can users boost higher, but for the same reason that bonding benefits the average speed, bonding makes it easier to boost big by siphoning a bit of bandwidth off of multiple channels.
It's because we're talking about the average case however that not everyone sees immediate benefits. Most users with PowerBoost available should see some kind of benefit, otherwise the biggest benefit will be for those users with a particularly unlucky channel allocated to them. And of course this assumes a DOCSIS 3.0 headend; none of these benefits would be seen with a DOCSIS 2.0 headend.
