Re: WAL write of full pages

You are correct, modern drives are much faster than this, for big, 
cacheable writes.

Try compiling and running the following code and watching your disk I/O.

Than, comment out the fsync(), which will make the writes cacheable.

Notice the huge difference.  It is an multiple of 15 difference on my 
machine.

#include <stdlib.h>
#include <stdio.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>

int main()
{        int i;        char    buf[8192];        int     ld;
        unlink("dump.out");        ld=open("dump.out", O_WRONLY | O_CREAT);
        for (i=0; i<65536; i++) {                write(ld, buf, sizeof(buf));                fsync(ld);
}
        close(ld);
        return 0;
}


Manfred Spraul wrote:
> Marty Scholes wrote:
> 
>>
>> 2. Put them on an actual (or mirrored actual) spindle
>> Pros:
>> * Keeps WAL and data file I/O separate
>> Cons:
>> * All of the non array drives are still slower than the array
> 
> 
> Are you sure this is a problem? The dbt-2 benchmarks from osdl run on an 
> 8-way Intel computer with several raid arrays distributed to 40 disks. 
> IIRC it generates around 1.5 MB wal logs per second - well withing the 
> capability of a single drive. My laptop can write around 10 MB/sec 
> (measured with dd if=/dev/zero of=fill and vmstat), fast drives should 
> be above 20 MB/sec.
> How much wal data is generated by large postgres setups? Are there any 
> setups that are limited by the wal logs.
> 
> -- 
>    Manfred
> 
> 
> 
> ---------------------------(end of broadcast)---------------------------
> TIP 6: Have you searched our list archives?
> 
>               http://archives.postgresql.org