N*****m 发帖数: 42603 | 1 你自己搞个server/cluster如果系统能auto-scale,资源elastic,那就是private clo
ud。问题是自己的private cloud如果用户量不大的话,要不浪费,要不会有瓶颈。 |
g*****g 发帖数: 34805 | 2 Not just elastic, high availability or so. You can easily have 2 regions, 6
zones in US with Amazon for availability. It's not possible to achieve that
unless you are Google, Facebook, M$ etc..
clo
【在 N*****m 的大作中提到】 : 你自己搞个server/cluster如果系统能auto-scale,资源elastic,那就是private clo : ud。问题是自己的private cloud如果用户量不大的话,要不浪费,要不会有瓶颈。
|
z*********e 发帖数: 10149 | 3 must pay big too...
6
that
【在 g*****g 的大作中提到】 : Not just elastic, high availability or so. You can easily have 2 regions, 6 : zones in US with Amazon for availability. It's not possible to achieve that : unless you are Google, Facebook, M$ etc.. : : clo
|
c*********e 发帖数: 16335 | 4 en,这个多zone非常贵,没有那么多公司喜欢把自己放别人手里控制。
【在 z*********e 的大作中提到】 : must pay big too... : : 6 : that
|
a******n 发帖数: 5925 | 5 disaster recovery 是另外一个
clo
【在 N*****m 的大作中提到】 : 你自己搞个server/cluster如果系统能auto-scale,资源elastic,那就是private clo : ud。问题是自己的private cloud如果用户量不大的话,要不浪费,要不会有瓶颈。
|
g*****g 发帖数: 34805 | 6 Not really, let's say you have a 3 nodes web cluster. You can make them in 3
zones in US-east, one each. Your service won't go down even if one zone is
on fire. You pay the same wether you have 3 nodes in 1 zone, or 3 nodes in
3 zones.
【在 z*********e 的大作中提到】 : must pay big too... : : 6 : that
|