We do a quarter yearly backup checks on our Postgresql 8.4 databases. We do this on VMWare virtual server that is not the latest in line of hardware.

The problem is with WAL-files. I'm restoring some 10k WAL files (thats about half of daily from last days full backup) and it takes me over 24 hours to restore them.

It looks like the virtual server is on economy mode. I'm using max 9 % of CPU allocated to the server. Also I/O speeds are very slow compared to for example unpacking the datafiles to the files system. Only thing that seems to be used is memory. (500 MB total of 512 MB). The server is not swapping. CPU usage avarege is 165 MHz. Average Disk Write rate is 3216 KBps. Swap Used 76K.

VMWare ESXi 3.5 is used for testing. No other Virtual machines running at the same time. CPU Athlon 64b 1,8 GHz. Memory 512 MB. Disk Controllers are SATA I.

Any ideas why this is so slow? This can't be just because of low RAM? I'm afraid that if this reflects to production machines recovery times, when I do really need to restore I'm pretty much in trouble.

So any ideas?