It's a good ideea?
No, your database would have zero reliability and zero trust. You'd also not be the first to do this, and neither the first to quickly get roasted pretty hard around here that your database is neither reliable or trustworthy.
I see the point. Yes, you are right about this.
Regarding the the bitcrack and vanitysearch there is a big problem using it. Normal users are seeing a speed and they are happy that the numbers are big, but the real problem is the process behind.
The authors was cutting from the process and increasing the speed, but forgot about this:
Endomorphism — What It Is and Why It Matters
The secp256k1 curve (Bitcoin) has a special mathematical property. If you compute one EC point k·G = (x, y), you get 5 more valid keypairs for almost free:
Check
Operation
Cost
k
Direct point
Full EC step
λk
(β·x mod p, y)
1 field multiply
λ²k
(β²·x mod p, y)
1 field multiply
-k
(x, -y mod p)
1 negation (free)
-λk
negate above
free
-λ²k
negate above
free
Cost ratio: 1 expensive EC step → 6 address checks. The extra 5 come from two cheap multiplications + negations.
Until puzzle 135 is ok to use it, but on the real BTC addresses is useless without Endomorphism. On my software version i added this option so when you search, you have all 6 addresses checked besided the versions from Github that check only 1.