Subject Re: [IB-Architect] UDF replacement: native shared libraries vs. J ava
Author Jason Chapman
> On the point of speed: Many have already pointed out how CPU and cache
> speed have leap ahead of I/O and disk speed. From my view the relatively
> small speed degradation in executing byte code is complete lost in the
> of doing even one disk I/O. For a database server this is byte code vs.
> native code difference would usually be insignificant.
That depends, we run with 1.6GB of db & 1gb of RAM. This means that a
significant part of the db is in memory, therefore execution speed of memory
based code is critical, especially if you were to be prcessing the UDF for
every line of a large select.