Re: Tweaking NFSv4 to be friendlier to hardware

New Message Reply About this list Date view Thread view Subject view Author view Attachment view

From: David Robinson (robinson@jetsun.eng.sun.com)
Date: 08/06/99-07:30:07 PM Z


Date: Fri, 6 Aug 1999 17:30:07 -0700 (PDT)
From: David Robinson <robinson@jetsun.eng.sun.com>
Message-Id: <199908070030.RAA05654@jetsun.eng.sun.com>
Subject: Re: Tweaking NFSv4 to be friendlier to hardware

> > 2) Add a length field preceding each component operation and component
> > response to allow them to be skipped over
> 
> Yes, for something like a DMA engine it would make it a bit
> easier to scan a compound request looking for juicy chunks
> of DMA-able data.  The downside is that it encumbers every
> compound op with an extra 4 bytes and makes XDR encoding
> more difficult.  Is the extra complexity really worthwhile ?

I find it interesting that you are willing to add an unnecessary
variable length tag but balk at a 4 byte length field.

Complex protocol processors are relatively rare, but there is
significant modern hardware that does checksumming and/or packet
routing based on offsets in packets.  Adding an explicit length
would greatly improve this sort processing as well as allow
concurrent decoding and possibly processing of requests.

	-David


New Message Reply About this list Date view Thread view Subject view Author view Attachment view

This archive was generated by hypermail 2.1.2 : 03/04/05-01:47:26 AM Z CST