At 09:09 �� 8/2/2002 +0000, you wrote:
>On Fri, 8 Feb 2002, Phoebus R. Dokos wrote: > >>Well I don't remember exactly what the problem is... but I think (correct >>me if I am wrong Claus) that Slave blocks are not cleared at all after >>being used and that effectively reduces the caching mechanism in constant >>streams of (new) data to 0... Well I think! > >Ok, that made little sense to *me*... > >Let's try this approach :o) > >What is a slave block, and how does it operate? I think that was answered... >What characteristic of them was good so they were written in in the first >place? Caching for Microdrives etc... >Why does streaming data cause problems for slave blocks? See Claus' answer... The memory used for caching for one is not reconfigurable and as Claus said the algorithm sucks (The Supervisor mode doesn't help either) >How could the problem be solved, WITH slave blocks? Change the Slaveing mechanism to use different caching algorithms and user defined cache size >How could the problem be solved, WITHOUT slave blocks? Remove them and you're ok although several problems need to be addressed: 1. Some form of caching is essential for today's applications 2. The QWA filesystem in itself is flawed so it needs to be fixed (to mention a few) >What implications are there for QDOS+ if there's no slave block(s)? Slow access (SOME form of caching is needed) >Thanks :o) > >Dave >ql.spodmail.com > >yw Phoebus
