Reducing cost: N + 1 spare and N + M spares scenarios
In the N + 1 spare and N + M spares scenarios, there are N "primary" servers, each of which operates in STANDALONE state, that is, without being connected to a secondary server. In addition, there are M spare servers, where M is at least 1 and usually less than N. If a "primary" server fails, one of the spares replaces it.
There are no secondary servers. A failed (primary) server is replaced with a spare server. The stand-alone server case is not a true hot standby scenario. It can be called warm standby; the spare server is available but it does not have an up-to-date copy of the database.
Data is copied from the original server to the spare, the original server is taken offline, and the spare is configured to act as the original server. Any spare can replace any primary server (no spare is dedicated to a particular primary server).
The N + 1 approach is referred to as single-spare scenario and N + M is referred to as multiple-spare scenario.
Both approaches require that you have a copy of the data of the original server somewhere. The single-spare and multiple-spare scenarios do not prevent data loss if the disk drive of the original server is damaged and there is no backup of the data.
The N+M approach can be useful in the following situations:
1 You are using the spare nodes to handle scheduled maintenance, not unexpected failures.
2 You have reliable backups that you can quickly copy to the spare server.
a You have backups on tape or on some other safe location.
b You are using solidDB Advanced Replication technology, and you can copy or re-create enough of the data by reading from the Advanced Replication master or replicas of the server that failed.
3 Individual pieces of data are not critical or are not unique.
a For example, if what you need is the "computing horsepower" (load-spreading capability) rather than the specific data, you might be able to meet your needs by copying a standard or "seed" database, or getting the data from clients, and then continue to run.
b Similarly, if all the servers have approximately the same data and are responding almost entirely to read requests with few or no write requests, you can copy a useful database from any one of your computers. For example, if you are running a large number of servers that all use the same Internet routing tables, or telephone directory information, you can use any of the server for recovery of data.