Third

6 views
Skip to first unread message

ambarisha b

unread,
Feb 11, 2011, 8:20:02 AM2/11/11
to asig-bi...@googlegroups.com
Well , anybody come up with anything for the third meet ? I have been
studying AFS but I guess this isn't the time to discuss that. So
somebody has to take the lead on this one.If any of you has something
to talk to about, I can try and read about it, so we can have a
discussion. In any case let's not skip the meet.

//Ambarish

Aastha Mehta

unread,
Feb 11, 2011, 8:31:27 AM2/11/11
to asig-bi...@googlegroups.com
What all file system types have you discussed so far?
How about distributed file system?
--
Aastha Mehta
Intern, NetApp, Bangalore
4th year undergraduate, BITS Pilani
E-mail: aast...@gmail.com


ambarisha b

unread,
Feb 11, 2011, 8:49:33 AM2/11/11
to asig-bi...@googlegroups.com
Yeah , AFS is a distributed file system. But we thought we should get
done with local filesystems before going on to distributed
implementations.Our idea was to study BSD and ext in feb and first
half of march , then move on to AFS and try and deploy AFS on a
personal network or if possible in the csa room during apogee.So, I
think jumping to that might hinder the study of local
filesystems.Anyway we planned to do it in mid march.

In the two meets we have had so far, we talked about basics of
filesystems and BSD's FFS.So, I think we still need to continue that.

Cheers
Ambarish

Aastha Mehta

unread,
Feb 11, 2011, 9:36:30 AM2/11/11
to asig-bi...@googlegroups.com
What concepts have been covered in file systems? Pardon my persisting question. I am just asking so that I can try suggest something new to read.

ambarisha b

unread,
Feb 11, 2011, 9:47:19 AM2/11/11
to asig-bi...@googlegroups.com
1 . Disk access model.
2 . File allocation policy
3 . Physical functioning of hard-disk and how that effects file
allocation policies.

Those , as a part of case study of FFS.

//Ambarish

kunreddy ramakrishna reddy

unread,
Feb 11, 2011, 10:50:03 AM2/11/11
to asig-bi...@googlegroups.com
@ambarisha i will try to discuss one possible solution -:) -:)  for the exercise which i posted in previous thread.Also will try to read FFS paper completely.Will bring one FFS university video(by some BSD expert) so that every one can see it and get some thing out of it(provided if we have good sound system).
--
Ramakrishna Reddy K, 
M.E. Computer Science,
Project Assistant,CAHU,BITS-Pilani,Rajasthan.

Aastha Mehta

unread,
Feb 11, 2011, 11:05:39 AM2/11/11
to asig-bi...@googlegroups.com
Cool. How about writing a file system of your own??
Yes, you heard it right :^)


There is a project called FUSE (File system in Userspace) which helps one to write their own file system completely in user space. What this means is, that you can implement a file system on top of the FUSE layer. While it itself runs on top of the conventional well designed file systems available, like ext2, ext3, VFS, etc.

FUSE itself also consists of user mode part and kernel mode part. One does not need to worry about the kernel mode part of it. The kernel mode part internally handles which file system underneath to map a syscall to. It provides a common single interface to a system call from each file system. For example, say there is a system call to write a file to disk, it maybe implemented differently in ext2 and ext3 with different interfaces as well. FUSE will provide a common write syscall interface for the user file system. Now, the user can create a file system on top of fuse and implement the write syscall as provided by FUSE. Based on kind of file to be written (whether it is located in an ext2 file system managed space or ext3, etc.), FUSE will multiplex the write call and call the syscall from appropriate filesystem underneath.

Well, there are tutorials available to write your own file system on top of FUSE:
1. http://www.cs.nmsu.edu/~pfeiffer/fuse-tutorial/
2. http://www.ibm.com/developerworks/linux/library/l-fuse/

I am working on writing a filesystem on top of FUSE myself right now. Let me know, if anyone's interested in this.

Aastha.

kunreddy ramakrishna reddy

unread,
Feb 11, 2011, 12:17:51 PM2/11/11
to asig-bitspilani
@Aastha  Quite interesting.First we will finish Local File Systems with understanding source code.After that we will try to deploy Open AFS file system and try to show demo at APOGEE.Than we will try FUSE.Thanks for posting its really useful.Can you share Architecture of that FUSE tool.

Aastha Mehta

unread,
Feb 11, 2011, 12:49:55 PM2/11/11
to asig-bi...@googlegroups.com
I came across this ppt. Gives a decent idea on how it works.
Introduction_to_FUSE.ppt

kunreddy ramakrishna reddy

unread,
Feb 13, 2011, 9:28:09 AM2/13/11
to asig-bi...@googlegroups.com
The paper attached contains one technique to improve Disk Bandwidth utilization which i discussed in yesterday's session.Just go through it.
lumb.pdf

UDBHAV PRASAD

unread,
Feb 13, 2011, 11:19:52 PM2/13/11
to asig-bi...@googlegroups.com
Can the timings for each meet be confirmed in the thread please? I was not aware that we had a session yesterday!! My fault, but I would be obliged if the timings were confirmed.

kunreddy ramakrishna reddy

unread,
Feb 14, 2011, 1:10:35 AM2/14/11
to asig-bi...@googlegroups.com
@Udbhav Prasad

Important Notice to all  ASIGians -:) -:)

Timings:
Every Saturday
8:00PM to 9:30PM
Venue: 1232 FD-I

If there is any change will be intimated before the session.

kunreddy ramakrishna reddy

unread,
Feb 14, 2011, 1:23:25 AM2/14/11
to asig-bi...@googlegroups.com
Exercise 2 (it may be wrong) : Needs to be done before coming to Session-4.Solution will be discussed in that session.

What is the relationship between "Virtual Memory" and "File System".If there is a relationship what is that?? is that File System normal(UFS,FFS etc..) one or special.If it is special then what it is and what are its characteristics ??
(Offcampus (&&) Oncampus ASIGians: Please modify the exercise if any thing wrong in that question and rephrase it properly.)

kunreddy ramakrishna reddy

unread,
Feb 14, 2011, 4:46:22 AM2/14/11
to asig-bitspilani
@Pranav the following might be the answer for your doubt.(Why available storage is less than free storage)

"Ext2 and higher FS reserves some blocks for the super user(root).Normally ,5% of the blocks are reserved.This allows the administrator to recover easily from situations where user processes fill up file systems".
Reply all
Reply to author
Forward
0 new messages