在MySQL中存储/检索IPv4和IPv6地址

Question

Sorry I don't know much about this subject, but all I'm looking for is a quick-and-easy solution for uniquely representing any IP address (v4/v6) in MySQL so I can easily retrieve the last time (if any) that a particular computer has visited my site.

I don't need to do any computations on the addresses, just retrieval, so any unique representation should be ok. I do plan on storing a lot of these (don't have an estimate yet), so space may become a concern.

I've seen many solutions for storing IP addresses, but it's unclear which work for both versions. MySQL's built-in INET_ATON doesn't seem to support IPv6. PHP's inet_pton seems promising but requires prior knowledge of the address's format. I'm also unsure about its usage (MySQL field type and writing the insertion statement via PHP). I've seen varchar(39) used to represent IPv6 addresses as strings, and I like that this solution is somewhat independent of server configuration; however, I'm a little uneasy about disk space. Would this approach be sufficient for all addresses that $_SERVER['HTTP_CLIENT_IP'] might output?

I'm a little surprised there isn't an obvious generic solution. I assumed this was a very common task. I'm having indecision about this single issue, and would like to move on with my project. Is a quick-and-easy solution unreasonable?

Thanks very much for any guidance...

Answer 1

我会这样：来自那里的citat： 如何在关系数据库中存储与IPv6兼容的地址 “最终决定：如果第二个bigint为NULL则为2xBIGINT，则表示IPv4”

Answer 2

It sounds like your main concern is about space. If that's the case, then you could use the fact that IPv4 addresses are (essentially) 32-bit numbers and IPv6 are 128. An IPv4 address can be stored in an INT column, but IPv6 would require two BIGINT columns in MySQL. This is likely to be much more space-efficient than storing strings.

The cost of doing this is that you need to do the conversion from address -> number before inserting the value into the database. That will (slightly) increase CPU load on your web server, so you need to figure out where your bottleneck is going to be and optimise for that.

An added benefit of storing the addresses as numbers is that you can have a very efficient index on the column(s) so looking up an address will be lightning fast. Indexing varchar columns is very expensive.