[Gluster-devel] Patch for "Striped" read from AFR volumes

[Csibra Gergo]

Monday, December 31, 2007, 7:14:24 PM, Krishna Srinivas wrote:

> The patch contribution is really appreciated. I did not verify the
> correctness of
> the code but I can make out that you are doing RR of readv().

Yes, you've right.

> But making read()s round-robin will decrease the performance (theoritically)
> as we wont be taking advantage of read ahead algorithm of the kernel.

Oh, I don't count with kernel read ahead, maybe yes.

> The better approach would be to make a file to be read from the same child
> everytime (even on the next open) but make different files to be read from
> different children. A good way of deciding the child to be read from is
> by (inode_number % child_count), this change is in the TLA repository.

I understand this...

> Could you test how your patch performs against the TLA source?

Yes I can test, I'll do this test in this week, and I'll send a patch
for TLA/1.3.8.

> A better way to define striped reads would be: if a read request comes for 1MB,
> get 0.5 MB from first child and 0.5MB from second child and combine the reads.

Yes this is the exact definition :)

Other: There's some misconfiguration in this mailing list. My email
address is in the "To:" field, and the gluster-develATnongluDOTorg in
the CC field. This means, If I make a Reply, I reply only to the
people who wrote the message, and not to the list. Must reply-all and
modify the "To:" field.

-- 
Best regards,
 Csibra Gergo                            mailto:gergo at csibra.hu