On Tue, Nov 25, 2003 at 01:57:33PM +0000, mb/ext3@xxxxxxxxxxxxxx wrote: > On Nov 22 Wen-han Goh wrote: > > >If I understand correctly, full journalling mode gives better > >performance for applications that do a lot of updates in O_SYNC. Could > >you please explain how this is possible? Doesn't full data journalling > >do twice as many writes as meta data journalling? > > Yes it does; however, once the data is in the journal the O_SYNC caller > can continue. With other modes you have to wait until the journal is > committed, and all the extra seeks that may ensue... More to the point, writes to the journal don't reqiure any seeks. So once the data has been written to the journal, you don't have to wait for the data to be written to their final locations on disk. This can result in faster times than not doing any journalling at all, because of the ability to reduce seeks. This is especially true if the journal is located on a separate block device, either on a separate disk spingle, or especially if it is a battery-backed up NVRAM device.... - Ted _______________________________________________ Ext3-users@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/ext3-users
