[OpenAFS] memcache option for Windows client

Rob Murawski rsm4@cssd.pitt.edu
Sun, 16 Nov 2003 19:39:48 -0500


This is a multi-part message in MIME format.

------=_NextPart_000_0007_01C3AC79.650EB9B0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

Re: [OpenAFS] memcache option for Windows clientSince the memory-mapped =
file is addressed as if it were memory, it should be fairly easy to =
change the startup code to malloc an appropriate amount of memory =
instead of creating the memory-mapped file.  I would presume no one has =
had the need to do this before.

-Rob
  ----- Original Message -----=20
  From: Christopher D. Clausen=20
  To: openafs-info@openafs.org=20
  Sent: Sunday, November 16, 2003 7:31 PM
  Subject: Re: [OpenAFS] memcache option for Windows client


  Rob Murawski <rsm4@cssd.pitt.edu> wrote:=20
  > An in-memory cache option doesn't make sense for the Windows client. =

  > It uses a memory mapped file for the cache.  This means that, if=20
  > sufficient memory exists, it is in memory.  It basically works like=20
  > virtual memory where it will be mapped back to the disk when needed, =

  > but is addressable like memory.=20
  >=20
  > Do you have a need that the cache ONLY be in memory?=20

  Actually, yes.=20

  I have some clients with largish amounts of RAM, low disk space, and=20
  slow disks.  It would be nice to avoid disk writes, as it =
significantly=20
  reduces performance.  And the cache size would not be dependant on the =

  amount of free disk space on the machine, but only the RAM.=20

  Perhaps it might be time to buy some larger, faster disks.=20

  I was also attempting to create a disk-less Windows client using =
Windows=20
  XP embedded, where disk access would complicate matters (although I =
have=20
  not gotten far yet.)=20

  <<CDC=20
  Christopher D. Clausen=20
  ACM@UIUC SysAdmin=20



  _______________________________________________=20
  OpenAFS-info mailing list=20
  OpenAFS-info@openafs.org=20
  https://lists.openafs.org/mailman/listinfo/openafs-info=20

------=_NextPart_000_0007_01C3AC79.650EB9B0
Content-Type: text/html;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD><TITLE>Re: [OpenAFS] memcache option for Windows =
client</TITLE>
<META http-equiv=3DContent-Type content=3D"text/html; =
charset=3Diso-8859-1">
<META content=3D"MSHTML 6.00.3790.94" name=3DGENERATOR>
<STYLE></STYLE>
</HEAD>
<BODY bgColor=3D#ffffff>
<DIV><FONT size=3D2>Since the memory-mapped file is addressed as if it =
were=20
memory, it should be fairly easy to change the startup code to malloc an =

appropriate amount of memory instead of creating the memory-mapped =
file.&nbsp; I=20
would presume no one has had the need to do this before.</FONT></DIV>
<DIV><FONT size=3D2></FONT>&nbsp;</DIV>
<DIV><FONT size=3D2>-Rob</FONT></DIV>
<BLOCKQUOTE dir=3Dltr=20
style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
  <DIV style=3D"FONT: 10pt arial">----- Original Message ----- </DIV>
  <DIV=20
  style=3D"BACKGROUND: #e4e4e4; FONT: 10pt arial; font-color: =
black"><B>From:</B>=20
  <A title=3Dcclausen@acm.org =
href=3D"mailto:cclausen@acm.org">Christopher D.=20
  Clausen</A> </DIV>
  <DIV style=3D"FONT: 10pt arial"><B>To:</B> <A =
title=3Dopenafs-info@openafs.org=20
  href=3D"mailto:openafs-info@openafs.org">openafs-info@openafs.org</A> =
</DIV>
  <DIV style=3D"FONT: 10pt arial"><B>Sent:</B> Sunday, November 16, 2003 =
7:31=20
  PM</DIV>
  <DIV style=3D"FONT: 10pt arial"><B>Subject:</B> Re: [OpenAFS] memcache =
option=20
  for Windows client</DIV>
  <DIV><FONT size=3D2></FONT><BR></DIV><!-- Converted from text/plain =
format -->
  <P><FONT size=3D2>Rob Murawski &lt;<A=20
  href=3D"mailto:rsm4@cssd.pitt.edu">rsm4@cssd.pitt.edu</A>&gt; =
wrote:</FONT>=20
  <BR><FONT size=3D2>&gt; An in-memory cache option doesn't make sense =
for the=20
  Windows client.</FONT> <BR><FONT size=3D2>&gt; It uses a memory mapped =
file for=20
  the cache.&nbsp; This means that, if</FONT> <BR><FONT size=3D2>&gt; =
sufficient=20
  memory exists, it is in memory.&nbsp; It basically works like</FONT> =
<BR><FONT=20
  size=3D2>&gt; virtual memory where it will be mapped back to the disk =
when=20
  needed,</FONT> <BR><FONT size=3D2>&gt; but is addressable like =
memory.</FONT>=20
  <BR><FONT size=3D2>&gt;</FONT> <BR><FONT size=3D2>&gt; Do you have a =
need that the=20
  cache ONLY be in memory?</FONT> </P>
  <P><FONT size=3D2>Actually, yes.</FONT> </P>
  <P><FONT size=3D2>I have some clients with largish amounts of RAM, low =
disk=20
  space, and</FONT> <BR><FONT size=3D2>slow disks.&nbsp; It would be =
nice to avoid=20
  disk writes, as it significantly</FONT> <BR><FONT size=3D2>reduces=20
  performance.&nbsp; And the cache size would not be dependant on =
the</FONT>=20
  <BR><FONT size=3D2>amount of free disk space on the machine, but only =
the=20
  RAM.</FONT> </P>
  <P><FONT size=3D2>Perhaps it might be time to buy some larger, faster=20
  disks.</FONT> </P>
  <P><FONT size=3D2>I was also attempting to create a disk-less Windows =
client=20
  using Windows</FONT> <BR><FONT size=3D2>XP embedded, where disk access =
would=20
  complicate matters (although I have</FONT> <BR><FONT size=3D2>not =
gotten far=20
  yet.)</FONT> </P>
  <P><FONT size=3D2>&lt;&lt;CDC</FONT> <BR><FONT size=3D2>Christopher D. =

  Clausen</FONT> <BR><FONT size=3D2>ACM@UIUC SysAdmin</FONT> </P><BR>
  <P><FONT =
size=3D2>_______________________________________________</FONT>=20
  <BR><FONT size=3D2>OpenAFS-info mailing list</FONT> <BR><FONT=20
  size=3D2>OpenAFS-info@openafs.org</FONT> <BR><FONT size=3D2><A=20
  =
href=3D"https://lists.openafs.org/mailman/listinfo/openafs-info">https://=
lists.openafs.org/mailman/listinfo/openafs-info</A></FONT>=20
  </P></BLOCKQUOTE></BODY></HTML>

------=_NextPart_000_0007_01C3AC79.650EB9B0--