«

Hi,

I am developing a C Program for reading over a million files of size 1
kilobytes each and sending the contents to another program using some
middle ware. I need some help on designing the program to process such
a large number of files in less than 8 hours.

TIA
Soren

On 15 Dec 2004 19:50:00 -0800, (soren juhu)
wrote in comp.lang.c:

> Hi,
>
> I am developing a C Program for reading over a million files of size 1
> kilobytes each and sending the contents to another program using some
> middle ware. I need some help on designing the program to process such
> a large number of files in less than 8 hours.
>
> TIA
> Soren

From the information you have provided in your post, the only advice
anyone could possibly give you would be to buy a faster computer with
faster hard disk drives to run your program on.

Even if you posted detailed information about the "processing" that
you had to do on the files, you don't have a C language question, you
have one about choosing the most efficient algorithm. For that you
need to post to an algorithm group such as news:comp.programming, and
be very explicit about the processing you need to do.

Once you have selected an algorithm, possibly with the help of an
appropriate group, if you have difficulties writing standard C code
that compiles and executes correctly, then post the problem code here,
explain your problems with it, and ask for C language advice.

--
Jack Klein
Home: http://JK-Technology.Com
FAQs for
comp.lang.c http://www.eskimo.com/~scs/C-faq/top.html
comp.lang.c++ http://www.parashift.com/c++-faq-lite/
alt.comp.lang.learn.c-c++
http://www.contrib.andrew.cmu.edu/~a...FAQ-acllc.html

soren juhu wrote:
>
> Hi,
>
> I am developing a C Program for reading over a million files of size 1
> kilobytes each and sending the contents to another program using some
> middle ware. I need some help on designing the program to process such
> a large number of files in less than 8 hours.

If we make the liberal assumption that you can locate and open each
file in 25 millisecs, that leaves you about 3800 seconds to process
1e9 bytes, or you will require a throughput in the order of 250k
bytes per second. Have fun.

--
Chuck F () ()
Available for consulting/temporary embedded and systems.
<http://cbfalconer.home.att.net> USE worldnet address!

On Wed, 15 Dec 2004 19:50:00 -0800, soren juhu wrote:

> Hi,
>
> I am developing a C Program for reading over a million files of size 1
> kilobytes each and sending the contents to another program using some
> middle ware. I need some help on designing the program to process such
> a large number of files in less than 8 hours.

Your main bottleneck is likely to be the file access. Accessing lots of
small files over a hard disk could end up being very slow, disks are much
more efficient reading large chunks of sequential data. You may want to
consider how your file os organised in the first place. If for example you
had the data written in 1K blocks in a single file (perhaps even do both)
the problem reduces to transferring a gigabyte of data which can be done
in seconds or minutes with normal LAN speeds.

This isn't a question about C but about the design of a system of
file management. You need to sit down and specify your real requirements,
e.g. why there are over a million 1K files in the first place and whether
a better approach is possible. There may be things you can do to aid this
transfer process when those million files are being generated (such as
append them to a single file, perhaps even put them in a database).

There is a lot you need to consider before worrying about C related issues.

Lawrence

soren juhu wrote:

> Hi,
>
> I am developing a C Program for reading over a million files of size 1
> kilobytes each and sending the contents to another program using some
> middle ware. I need some help on designing the program to process such
> a large number of files in less than 8 hours.

Do the arithmetic.

1000000 * 1024 * 8 (assuming an 8-bit-byte platform for the moment)
comes to 8192000000 bits. If you have, say, 7 hours to transfer this
amount of data, you will need to throw bits down the wire at a
rate of at least 325 kbps. This should easily be within the reach
of modern network cards. I don't think you'll have a problem.

I suggest you write an "obvious" program, and then test it to
see if it's quick enough. If so, fabulous. If not, post it here
and maybe we can help you speed it up.

It's worth remembering that this newsgroup can't - or rather,
won't - help you on the networking aspects of such a program.
But they are likely to come up with some good ideas on the
rest of it, given the catalyst of some source code to inspect.

Best of luck.

soren juhu wrote:
> Hi,
>
> I am developing a C Program for reading over a million files of size
1
> kilobytes each and sending the contents to another program using some
> middle ware. I need some help on designing the program to process
such
> a large number of files in less than 8 hours.
>
> TIA
> Soren


Hi all,

Sorry for my late reply, I am posting my message using Google Groups.

Thanks a lot for your valuable inputs to the problem. It definitely
helped in knowing where to start for solving the problem. I will surely
inform you about this development effort.

Thanks,
Soren

In article <>,
Jack Klein <> wrote:
>On 15 Dec 2004 19:50:00 -0800, (soren juhu)
>wrote in comp.lang.c:
>
>> Hi,
>>
>> I am developing a C Program for reading over a million files of size 1
>> kilobytes each and sending the contents to another program using some
>> middle ware. I need some help on designing the program to process such
>> a large number of files in less than 8 hours.
>>
>> TIA
>> Soren
>
>From the information you have provided in your post, the only advice
>anyone could possibly give you would be to buy a faster computer with
>faster hard disk drives to run your program on.
>
>Even if you posted detailed information about the "processing" that
>you had to do on the files, you don't have a C language question, you
>have one about choosing the most efficient algorithm. For that you
>need to post to an algorithm group such as news:comp.programming, and
>be very explicit about the processing you need to do.

I interpret the question this way: what standard c-function
are appropriate and how should I use them.

My answer proves you wrong.

1. Assuming you can guarantee a maximum size of each file,
read them in one go in a static buffer of that size.
2. Go for the lowest level calls, (read/write) and
handle the rest yourself.
3. You total througput seems to be in reach for modern
disks. C is low overhead and shouldn't get in your
way for a reasonable amount of processing.

There is no way the OP ask about "processing to be done".

>Once you have selected an algorithm, possibly with the help of an
>appropriate group, if you have difficulties writing standard C code

There is no way you could mention an algorithm. You have not the
slightest clue, if you wanted to. The OP was well aware that
would be off topic.

>that compiles and executes correctly, then post the problem code here,
>explain your problems with it, and ask for C language advice.

Aren't we going overboard? Such that only home work questions
are appropriate?

>--
>Jack Klein

Groetjes Albert.

--
--
Albert van der Horst,Oranjestr 8,3511 RA UTRECHT,THE NETHERLANDS
One man-hour to invent,
One man-week to implement,
One lawyer-year to patent.