From: SMTP%"ntfsd@atria.com" 10-DEC-1997 16:20:57.51 To: ntfsd-digest@atria.com CC: Subj: [ntfsd] ntfsd-digest V1 #270 Return-Path: owner-ntfsd@atria.com Received: by arisia.gce.com (UCX V4.1-12C, OpenVMS V7.1 VAX); Wed, 10 Dec 1997 16:15:42 -0500 Received: from gw.atria.com (gw.atria.com [192.88.237.2]) by mercury.mv.net (8.8.8/mem-971025) with SMTP id HAA14321 for ; Wed, 10 Dec 1997 07:20:54 -0500 (EST) Received: by gw.atria.com id Wed, 10 Dec 1997 04:11:16 -0500 Received: from milan.doe.ernet.in by gw.atria.com id Wed, 10 Dec 1997 04:11:08 -0500 Received: from sac.ernet.in ([202.141.100.40]) by milan.doe.ernet.in (4.1/SMI-4.1) id AA03441; Wed, 10 Dec 97 14:36:45+050 Received: from gw.atria.com ([192.88.237.2]) by sac.ernet.in (4.1/SMI-4.1) id AA06480; Wed, 10 Dec 97 14:43:40 HKT Received: by gw.atria.com id Wed, 10 Dec 1997 01:00:53 -0500 Date: Wed, 10 Dec 1997 01:00:53 -0500 Message-Id: <199712100600.BAA16252@gw.atria.com> From: owner-ntfsd-digest@atria.com (ntfsd-digest) To: ntfsd-digest@atria.com Subject: [ntfsd] ntfsd-digest V1 #270 Reply-To: ntfsd@atria.com Sender: owner-ntfsd@atria.com Precedence: bulk ntfsd-digest Wednesday, December 10 1997 Volume 01 : Number 270 ---------------------------------------------------------------------- Date: Tue, 9 Dec 1997 17:41:50 -0800 From: Marc Kupper Subject: RE: [ntfsd] FAT12 & FAT16 > -----Original Message----- > From: Alexander E. Sobol [SMTP:mlist@mccme.rssi.ru] > Sent: Monday, December 08, 1997 6:34 PM > To: Marc Kupper > Subject: RE: [ntfsd] FAT12 & FAT16 > > Thank you for your reply; > that actally helped me; > My will create partitions with 1 sector per cluter ratio; > (> 2Mb, as you pinted). > > BTW, why do they usually use larger cluster sized ? > I assume it will lead only to disk space loss. > (at least halp a cluster per file - in average, and much more for > small files) [NTFSD list - I tried to send this response directly back to Sasha and am getting errors back from my mailer. We just upgraded to Exchange 5.5 and I think it has a bug where if an MX record for A.B.C points to X.A.B.C then Exchange gets upset.] The reason for going to a larger and larger cluster sizes is that 16-bit FAT partitions can have from 4080 to 65519 clusters. (see below for exactly why I used these numbers). If the cluster size is 32K bytes, a common size, then the largest partition you can support is 2 gigabytes. To install one FAT partition on a 3 gigabyte drive (about the smallest you can buy now) you would need to go to 64K bytes/cluster and would have about 46,000 clusters. Yes, it can waste a LOT of hard disk space if most of your files are well under 64K bytes long. This is why FAT-32 was invented. That allows up to 4 billion clusters. Now you could set up your 3 gigabyte hard disk with 1.5 million 2K byte clusters. Of course, the FAT balloons to 6 megabytes per copy or 2% of the disk per copy. A more reasonable choice may be 4K or 8K clusters which would result in smaller FATs. I think that answers your question on why people are using large clusters. Here are the numbers on FAT sizes. A 12-bit FAT uses the following special values to mark entries in the FAT table which has one entry per cluster. 000 - Cluster is available for allocation 001 to FEF - Pointer to next cluster for a file FF0 to FF6 - Reserved cluster (Used to mark things such as the part of the disk covered by the FAT itself) FF7 - Bad spot on the hard disk FF8 to FFF - Last cluster of a file (you will almost always see FF8 or FFF) Thus the cluster numbers for files on a 12-bit partition can range from 001 to FEF or 1 to 4079 (decimal). 16-bit clusters use a very similar numbering system. 0000 - Cluster is available for allocation 0001 to FFEF - Pointer to next cluster for a file FFF0 to FFF6 - Reserved cluster FFF7 - Bad spot on the hard disk FFF8 to FFFF - Last cluster of a file (you will almost always see FFF8 or FFFF) Thus the cluster numbers for files on a 16-bit partition can range from 0001 to FFEF or 1 to 65,519 (decimal). If a hard disk will have from 1 to 4079 clusters then FORMAT sets the up the partition's BIOS Parameter Block (BPB) and FAT to be for 12-bit per FAT entry. If the disk has from 4080 to 65,519 clusters then FORMAT sets up a 16-bit BPB and FAT. I am kind of surprised this was not extended to include 20, 24, 28, and then 32 bit FAT partitions. Instead Microsoft jumped to FAT-32 and would guess the cluster tagging system is identical to what is shown above. Marc - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - [ To unsubscribe, send email to ntfsd-request@atria.com with body UNSUBSCRIBE (the subject is ignored). ] ------------------------------ Date: Tue, 9 Dec 1997 18:09:49 -0800 From: Daniel Lovinger Subject: RE: [ntfsd] FAT12 & FAT16 > -----Original Message----- > From: Marc Kupper [SMTP:MKupper@KentConsulting.com] > Sent: Tuesday, December 09, 1997 5:42 PM > To: Alexander E. Sobol; ntfsd@atria.com > Subject: RE: [ntfsd] FAT12 & FAT16 > > This is why FAT-32 was invented. That allows up to 4 billion clusters. > Actually, the top 4 bits of the index are reserved (all implementations are to be "glass" with respect to them), which limits FAT32 in practice to 2^28 =~ 268 million. Still, this is obviously far more than anyone would ever want to use. > I am kind of surprised this was not extended to include 20, 24, 28, and > then 32 bit FAT partitions. Instead Microsoft jumped to FAT-32 and > would guess the cluster tagging system is identical to what is shown > above. > Modulo that it is really "FAT-28", this is true. Growing the index in nibble-sized chunks really wouldn't serve to do anything but compress the FAT by less than a factor of two as compared to the theoretical FAT16 equivalent. This would also introduce issues with respect to mapping FATs through caches since individual index entries in the FAT could straddle page and NT cache manager view boundaries, and require extra synchronization since the adjacent indices would not line up on units capable of atomic writes on current machines. - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - [ To unsubscribe, send email to ntfsd-request@atria.com with body UNSUBSCRIBE (the subject is ignored). ] ------------------------------ Date: Wed, 10 Dec 1997 11:17:03 +0000 From: "Venkateswaran K." Subject: [ntfsd] FAT-32 File System Can any one tell me where I can find information on FAT-32 File System internals? Thanks in advance. With regards, Venkateswaran K Shuttle Technology (India) Private Limited, PO Box 225, PONDICHERRY, INDIA E-MAIL: venkat@shuttletech.co.in URL: http://www.shuttletech.com - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - [ To unsubscribe, send email to ntfsd-request@atria.com with body UNSUBSCRIBE (the subject is ignored). ] ------------------------------ End of ntfsd-digest V1 #270 *************************** - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - [ To unsubscribe, send email to ntfsd-request@atria.com with body UNSUBSCRIBE (the subject is ignored). ]