T
timeOday
I am making a system which uses a SQL database (currently H2) for data
acquisition. This requires many very small disk writes.
Initial benchmarking of the database with a plain-Jane 7200 RPM IDE
desktop drive, probably 3 or 4 years old, reveals the maximum rate of
insertions is only about 30% of what we need. It is I/O rather than CPU
bound.
We could put more effort ($$$) making the code more optimized (and
complex), but I'd rather just buy faster storage if possible.
Is a current fast hard drive (say, a Raptor) likely to triple database
insert speed? I'm guessing not. How about a RAID array - what sort of
card, which RAID level, and how many drives might be required?
I realize nothing is guaranteed based on the information I've provided,
but economically, it's well worth it for me to just buy it and see, if
there's a decent chance of success.
acquisition. This requires many very small disk writes.
Initial benchmarking of the database with a plain-Jane 7200 RPM IDE
desktop drive, probably 3 or 4 years old, reveals the maximum rate of
insertions is only about 30% of what we need. It is I/O rather than CPU
bound.
We could put more effort ($$$) making the code more optimized (and
complex), but I'd rather just buy faster storage if possible.
Is a current fast hard drive (say, a Raptor) likely to triple database
insert speed? I'm guessing not. How about a RAID array - what sort of
card, which RAID level, and how many drives might be required?
I realize nothing is guaranteed based on the information I've provided,
but economically, it's well worth it for me to just buy it and see, if
there's a decent chance of success.