Processing Scheduling Algorithm?

luke_church

Hi guys,

I've just uploaded three galleries worth of photos (~600Mb), here:

http://www.lukechurchphotography.com/Events/171312

About 70% of the photos were finished within 5 minutes of completing the upload :

About 90% were done within 10 minutes or so : :

The rest still aren't done, 2 hours later... :scratch

This is kind of irritating as it means that I have to delay the release fo the galleries, but I appreciate the difficulties of estimating timing for a statistical delay.

My question is then, can you give me any insight into the nature of your scheduling algorithms? Is there a 'cliff-edge' drop off if a task runs over a given limit? If so is a fixed penalty?

Is there anything I can do to estimate rough time boundaries or to make the process more deterministic? Or at least give me some insight into why the radically different delays. I can't imagine that simple FIFO would result in this...

Any insight into the process would be apprecaited...

I guess it may well be so complex that any assumptions are invalid, esp. if there is some kind of throttling going on...

Cheers,

Luke

onethumb

luke_church wrote:

Hi guys,

I've just uploaded three galleries worth of photos (~600Mb), here:

http://www.lukechurchphotography.com/Events/171312

About 70% of the photos were finished within 5 minutes of completing the upload :

About 90% were done within 10 minutes or so : :

The rest still aren't done, 2 hours later... <img src="https://us.v-cdn.net/6029383/emoji/headscratch.gif&quot; border="0" alt="" >

This is kind of irritating as it means that I have to delay the release fo the galleries, but I appreciate the difficulties of estimating timing for a statistical delay.

My question is then, can you give me any insight into the nature of your scheduling algorithms? Is there a 'cliff-edge' drop off if a task runs over a given limit? If so is a fixed penalty?

Is there anything I can do to estimate rough time boundaries or to make the process more deterministic? Or at least give me some insight into why the radically different delays. I can't imagine that simple FIFO would result in this...

Any insight into the process would be apprecaited...

I guess it may well be so complex that any assumptions are invalid, esp. if there is some kind of throttling going on...

Cheers,

Luke

FIFO.

Don

luke_church

onethumb wrote:

FIFO.

Don

Cheers Don.

It looks like something kind of odd was going on... Sorted now.

Thanks,

Luke