How likely are two data blocks that can lead to the same CRC64 value?

Question

I have a caching application that uses a CRC64 value to ensure data integrity. I’m thinking about putting an extra field, timestamps that will be transferred with data between different cache servers and compare them to see if the data has changed.

However, this requires protocol changes. Although this is not a huge deal, I already have CRC64, which can be used as an indicator that something has changed.

Does anyone know the statistics around two data blocks creating the same CRC64? If not, how can I calculate it or evaluate its probability?

Answer 1

, crc64 "", :

1% 6.1 × 10 ^ 8 . 50% 5.1 × 10 ^ 9 .

, , , , crc64, , . , , , .

Answer 2

1/2 ⁶⁴ 1 1,8 ; 10 ¹⁹.

, N.

. , .

Answer 3

, CRC64 , 2 ** 64. CRC , , . , , , , , , 1 2 ** 50 .

You would be sure to get close to the theoretical limit if you use a cryptographic hash instead of CRC64, but a cryptographic hash is usually much more expensive to calculate.