I don't know of anyone that is interesting on working on that specific feature. What we have been talking about is an automatic migration of objects up and down the storage  hierarchy.

Having it write to SSD first could be both good and bad (fast but lots of write wear).

On Mar 21, 2016, at 12:19 PM, Muhammad Faisal <faisalusuf@yahoo.com> wrote:

OK. But this feature can be added right? May be in future releases. It can improve caching performance by avoiding seek time of disks which increase over the period of time with high disk WR.
--
Regards,
Faisal.
 
 
 
------ Original Message ------
From: "Leif Hedstrom" <zwoop@apache.org>
To: users@trafficserver.apache.org; "Muhammad Faisal" <faisalusuf@yahoo.com>
Sent: 3/21/2016 9:52:35 PM
Subject: Re: Different Cache Disk for different size of objects
 

On Mar 18, 2016, at 3:35 AM, Muhammad Faisal <faisalusuf@yahoo.com> wrote:

Hi,
Is this possible to allocate a different disk for different object sizes? Below is the scenario I'm trying to implement:
 
Object size <1MB -----> SSD 
Object Size > 1MB ----> HDD
 
This will improve Caching performance as smaller objects will be served from SSD while larger objects will reside on the HDD.


No, not at this point. Part of the issue is that we select “storage” before going to origin, so you don’t know what the size is going to be before you get the response. And at that point, you (currently) can’t move to a different storage / volume. But @amc would know best.

— leif