Hello!
I've been wondering if range requests are more efficient than doing "whole" requests for relatively large objects (100MB-1GB).
More precisely, my doubt is regarding the use of OSD/RGW resources, that is, does the entire object is retrieved from the OSD only to be sliced afterwards? Or only the requested portion is read/sent from the OSD to the RGW?
The reason is that, in my scenario, the entire object may be requested to ceph eventually, either via multiple range requests or a single request.
But, from my application point of view, it would be more efficient to retrieve such object partially as needed, although only if such range requests do not end up using more resources than necessary from my ceph cluster (such as retrieving the whole object for each range request).
I've searched the online documentation, as well as the mailing list, but failed to find any indicative of how range requests are processed by ceph.
Thanks in advance.
Giovani.
_______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
