network redundancy

Network redundancy generally refers to the redundancy of network paths, that is, when one path (physical link) in the network fails and is disconnected, information can also be transmitted through other paths (physical links).
The landlord uses two servers, maybe it's like realizing data redundancy?
In fact, the easiest way to achieve data redundancy is to use a disk array . General servers support RAID0.
For applications, when there is only one server, if it goes down, the service will be interrupted.
If you want to recover in a short time, you need to see how much time affects the failure in practice. If it is very huge, you must have a solution.
If the time to solve the problem is not very urgent, for example, it is required that the service must be restored within an hour or two when the server is down. My suggestion is: the two servers share the same disk array, and only data (such as database) is placed in the array. data files), and things other than data (such as: operating system system files, some software, database system files) must be installed on the server, and the two servers should be set as much as possible (user name, password, IP address are not the same). Then, usually only one server is started. When one server goes down, we artificially discover that the service has stopped, and then another server can be started.
If the time to solve the problem is very urgent, and the service must be restored within one or two minutes, then the dual -system hot backup technology must be adopted, which is too costly.


Redundancy depends on the requirements, what do you want redundant, that is, what you are most worried about.
According to the meaning of the landlord, there seems to be only one server, which means that you are preparing a redundant network. If it is only a redundant network, you do not need dual network cards (because the addresses of dual network cards are not easy to configure, unless you configure a floating address), the standard approach is so:
The server network card is connected to a switch or HUB, and the two cables from the switch or HUB are respectively connected to your two gigabit switches. VRRP (Huawei) or HSRP (CISCO) is connected to the two gigabit switches, and a floating address can be configured.

However, in my experience, servers are often more prone to problems than networks, and perhaps server redundancy should be considered more.