| *** gkadam has joined #openstack-swift | 12:30 | |
| *** gkadam has quit IRC | 12:34 | |
| *** silor has joined #openstack-swift | 12:55 | |
| *** bhar12345 has joined #openstack-swift | 14:01 | |
| bhar12345 | Why are green drives not recommended in swift? I know that the power up and down can cause excess wear in the data but could someone elaborate more on it? This is very central to my research, it would help me a lot if someone could elaborate on this. | 14:02 |
|---|---|---|
| bhar12345 | This answer could make or break my research! | 14:03 |
| DHE | exactly that. the drives are rated for a number of load/unload cycles and greens go through them excessively fast. | 14:03 |
| DHE | greens probably unload after a mere 5-10 seconds of idle time | 14:03 |
| DHE | also other types of drives are better suited to operating in large groups of other drives, have longer warranties, and generally better endurance. | 14:04 |
| bhar12345 | ok ok | 14:11 |
| bhar12345 | So here is my research idea. In swift, We write the object to 3 disks. The idea is to keep one of the 3 disks in power saving mode and write the 3rd replica to an ssd. In this process we are aiming to reduce the power consumption. | 14:12 |
| bhar12345 | The 3rd disk which was in power saving mode will be powered on every 2-3 hours and shift the 3rd replica from the ssd to the 3rd disk. | 14:12 |
| bhar12345 | What do you guys think of this idea? | 14:12 |
| DHE | each other object storage server is going to be checking to see if this spun down drive has all the data it should. you'll have to convince them without spinning up the drive | 14:16 |
| bhar12345 | if they are not convinced they will write the object to the handoff nodes right | 14:18 |
| DHE | yes | 14:18 |
| bhar12345 | which is counter-productive. | 14:19 |
| bhar12345 | is it the case that people never use green drives in large scale swift deployments? | 14:19 |
| DHE | well I certainly can't speak to the whole world... but I wouldn't use them for anything honestly | 14:20 |
| bhar12345 | what do you think of the idea? is something like that implementable? if so will it help reduce power in a swift deployment? Would love to hear your 2 cents on this :) | 14:22 |
| *** mahatic has quit IRC | 14:23 | |
| bhar12345 | k | 14:53 |
| bhar12345 | kkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkk | 14:53 |
| bhar12345 | sorry! it was by accident! | 14:54 |
| bhar12345 | But i would love to hear all of your opinions on the problem: So here is my research idea. In swift, We write the object to 3 disks. The idea is to keep one of the 3 disks in power saving mode and write the 3rd replica to an ssd. In this process we are aiming to reduce the power consumption.The 3rd disk which was in power saving mode will be powered on every 2-3 hours and shift the 3rd replica from the ssd to the 3rd disk | 14:56 |
| DHE | well as-is swift can give you the guaranteed 3rd replica on an SSD. that's fairly straight-forward. even set it up to serve the content as much as possible over the spinning disks | 14:58 |
| *** bhar12345 has quit IRC | 15:05 | |
| *** bhar12345 has joined #openstack-swift | 15:18 | |
| *** bhar12345 has quit IRC | 15:18 | |
| *** bhar12345 has joined #openstack-swift | 15:23 | |
| *** silor has quit IRC | 15:45 | |
| *** silor has joined #openstack-swift | 15:49 | |
| bhar12345 | i would love to hear all of your opinions on the problem: So here is my research idea. In swift, We write the object to 3 disks. The idea is to keep one of the 3 disks in power saving mode and write the 3rd replica to an ssd. In this process we are aiming to reduce the power consumption.The 3rd disk which was in power saving mode will be powered on every 2-3 hours and shift the 3rd replica from the ssd to the 3rd disk | 15:58 |
| bhar12345 | the ssd is a cache for the 3rd replica. We will be putting the 3rd replica back to the 3rd primary node once it powered back on | 15:59 |
| *** silor has quit IRC | 16:18 | |
| *** bhar12345 has quit IRC | 16:21 | |
| *** tristanC has quit IRC | 16:24 | |
| *** MooingLemur has quit IRC | 16:24 | |
| *** timburke has quit IRC | 16:24 | |
| *** jarbod has quit IRC | 16:24 | |
| *** aluria has quit IRC | 16:24 | |
| *** mattoliverau has quit IRC | 16:24 | |
| *** timburke has joined #openstack-swift | 16:27 | |
| *** ChanServ sets mode: +v timburke | 16:28 | |
| *** Miouge has quit IRC | 16:28 | |
| *** tristanC has joined #openstack-swift | 16:28 | |
| *** bhar12345 has joined #openstack-swift | 16:29 | |
| *** Miouge has joined #openstack-swift | 16:31 | |
| *** silor has joined #openstack-swift | 17:28 | |
| *** MooingLemur has joined #openstack-swift | 18:20 | |
| *** bhar12345 has quit IRC | 18:41 | |
| *** awkwardpenguin has joined #openstack-swift | 19:16 | |
| *** pcaruana has quit IRC | 19:21 | |
| *** silor has quit IRC | 19:28 | |
| *** silor has joined #openstack-swift | 19:29 | |
| *** awkwardp_ has joined #openstack-swift | 19:29 | |
| *** awkwardpenguin has quit IRC | 19:32 | |
| *** awkwardp_ has quit IRC | 19:45 | |
| *** awkwardpenguin has joined #openstack-swift | 19:45 | |
| *** awkwardp_ has joined #openstack-swift | 19:45 | |
| *** awkwardpenguin has quit IRC | 19:49 | |
| *** awkwardp_ has quit IRC | 19:59 | |
| *** silor has quit IRC | 20:13 | |
| *** awkwardpenguin has joined #openstack-swift | 22:00 | |
| *** rcernin has joined #openstack-swift | 22:34 | |
| *** awkwardpenguin has quit IRC | 23:04 | |
| *** mattoliverau has joined #openstack-swift | 23:07 | |
| *** ChanServ sets mode: +v mattoliverau | 23:07 | |
| mattoliverau | morning | 23:07 |
| *** awkwardpenguin has joined #openstack-swift | 23:15 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!