Re: How about implementing an online hierarchical storage on ext2/ext3 file system?

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Sandeep K Sinha wrote:
Hey Scott,
What I can recall and and IMHO,
here online means thats the relocation of files from tier to the tier
should not affect the throughput of the system and and should not be
done after making the filesystem offline for use.

Also, Hierarchical here means the different storage classes based on
the quality and cost of the storage.
Something when you have a LVM on top of a collection of different
storage class storage.

Rohit, can you just confirm, if thats true.

Btw, Rohit can you brief on what exactly are your intentions.

On Sun, Jul 27, 2008 at 5:04 PM, rohit vashist
<rohitvashist2kk3@xxxxxxxxx> wrote:
hello

by online hierarchical i mean that during runtme can we decide a
placement policy
to place a file accordingly on the physical disk.i mean giving a
multiple tier soluntion
to the file for eg. file of great importance on tier1 of leasr
importance on tier2...

Whether such thing exits in ext2/ext3??

I don't believe that ext2/3 implement such a thing. Usually you get 'write behind', 'read ahead' for performance. IOW, you write to a free block on a track that you're going to do a read from to avoid costly head seeks, and downplay rotational latency. You can do what's called 'short stroking' your disk for performance. This is when you create 'bands' for data (through your partition scheme) such that you place all data that will be accessed together close to each other and on a track that makes sense for its speed requirements. For instance, my usual partitioning scheme is that /var and /usr go towards the outside of the disk (the first tracks, which are the fastest), swap goes in the middle such that worst case is a half disk seek, and static files are on the final tracks.

However, you can set priorities for swap file partitions/files. If two swaps have the same priority, they will be used in a round robin style. Usually though, the layout of the disk will mandate the placement of data. For instance, in software RAID the device used usually has to be a certain device. Like striped data interleaves every X bytes. In that case, the file system isn't aware of the underlying device (usually - some are and will match their strides to it, XFS will IIRC).

Does this answer your question, or did I misunderstand what you were asking? I tend to go off on tangents only to find out I've covered everything EXCEPT what was being asked ;) I've got bad S/N ratio at times :)

--
To unsubscribe from this list: send an email with
"unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx
Please read the FAQ at http://kernelnewbies.org/FAQ


[Index of Archives]     [Newbies FAQ]     [Linux Kernel Mentors]     [Linux Kernel Development]     [IETF Annouce]     [Git]     [Networking]     [Security]     [Bugtraq]     [Yosemite]     [MIPS Linux]     [ARM Linux]     [Linux RAID]     [Linux SCSI]     [Linux ACPI]
  Powered by Linux