Re: [Openstack] [metering] delta vs. cumulative meters
On Tue, May 29 2012, Doug Hellmann wrote: […] Thoughts? As you pointed out, it's not safe to compute delta when your counter is gauge, and vice versa. Rather than documenting, we may just add the counter type (à la RRD: gauge, counter, derive…) to the meter, so it's easier to make things evolve and discover counters by themselves. -- Julien Danjou // eNovance http://enovance.com // ✉ julien.dan...@enovance.com ☎ +33 1 49 70 99 81 ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp
Re: [Openstack] [metering] delta vs. cumulative meters
On Wed, May 30, 2012 at 4:18 AM, Julien Danjou julien.dan...@enovance.comwrote: On Tue, May 29 2012, Doug Hellmann wrote: […] Thoughts? As you pointed out, it's not safe to compute delta when your counter is gauge, and vice versa. Rather than documenting, we may just add the counter type (à la RRD: gauge, counter, derive…) to the meter, so it's easier to make things evolve and discover counters by themselves. I like that. I'll open a bug to do that. Doug ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp
[Openstack] [metering] delta vs. cumulative meters
IIRC, the meters discussed in the wiki [1] are supposed to show delta values (usage since the last time an event was generated), although the Alternate Gauge Design section discusses cumulative meters instead. The libvirt pollsters we have now produce cumulative data, and it might be complicated and error-prone to change them to compute the deltas internally before generating events. On the other hand, some of the other counters may be more complicated to generate as cumulative values. Rather than taking an either/or approach, I propose we document for each counter whether it uses the delta or cumulative approach. The API server can ask the meter plugin for that piece of information when calculating the aggregate value so that the caller does not have to keep track of the rules for each meter. Thoughts? Doug [1] http://wiki.openstack.org/EfficientMetering ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp
Re: [Openstack] [metering] delta vs. cumulative meters
On 05/29/2012 05:42 PM, Doug Hellmann wrote: IIRC, the meters discussed in the wiki [1] are supposed to show delta values (usage since the last time an event was generated), although the Alternate Gauge Design section discusses cumulative meters instead. The libvirt pollsters we have now produce cumulative data, and it might be complicated and error-prone to change them to compute the deltas internally before generating events. On the other hand, some of the other counters may be more complicated to generate as cumulative values. Rather than taking an either/or approach, I propose we document for each counter whether it uses the delta or cumulative approach. The API server can ask the meter plugin for that piece of information when calculating the aggregate value so that the caller does not have to keep track of the rules for each meter. I tend to prefer using a gauge and I agree that documenting what was preferred (gauge or delta) for a given meter / counter is enough. My 2ct ;-) -- Loïc Dachary Chief Research Officer // eNovance labs http://labs.enovance.com // ✉ l...@enovance.com ☎ +33 1 49 70 99 82 ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp
Re: [Openstack] [metering] delta vs. cumulative meters
Do we want a history of deltas or just last delta and leave the history to UI implementers? On Tue, May 29, 2012 at 9:30 AM, Loic Dachary l...@enovance.com wrote: On 05/29/2012 05:42 PM, Doug Hellmann wrote: IIRC, the meters discussed in the wiki [1] are supposed to show delta values (usage since the last time an event was generated), although the Alternate Gauge Design section discusses cumulative meters instead. The libvirt pollsters we have now produce cumulative data, and it might be complicated and error-prone to change them to compute the deltas internally before generating events. On the other hand, some of the other counters may be more complicated to generate as cumulative values. Rather than taking an either/or approach, I propose we document for each counter whether it uses the delta or cumulative approach. The API server can ask the meter plugin for that piece of information when calculating the aggregate value so that the caller does not have to keep track of the rules for each meter. I tend to prefer using a gauge and I agree that documenting what was preferred (gauge or delta) for a given meter / counter is enough. My 2ct ;-) -- Loïc Dachary Chief Research Officer // eNovance labs http://labs.enovance.com // ✉ l...@enovance.com ☎ +33 1 49 70 99 82 ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp
Re: [Openstack] [metering] delta vs. cumulative meters
On Tue, May 29, 2012 at 12:30 PM, Loic Dachary l...@enovance.com wrote: On 05/29/2012 05:42 PM, Doug Hellmann wrote: IIRC, the meters discussed in the wiki [1] are supposed to show delta values (usage since the last time an event was generated), although the Alternate Gauge Design section discusses cumulative meters instead. The libvirt pollsters we have now produce cumulative data, and it might be complicated and error-prone to change them to compute the deltas internally before generating events. On the other hand, some of the other counters may be more complicated to generate as cumulative values. Rather than taking an either/or approach, I propose we document for each counter whether it uses the delta or cumulative approach. The API server can ask the meter plugin for that piece of information when calculating the aggregate value so that the caller does not have to keep track of the rules for each meter. I tend to prefer using a gauge and I agree that documenting what was preferred (gauge or delta) for a given meter / counter is enough. Just to be clear, by gauge do you mean the cumulative form of the counter? My 2ct ;-) -- Loïc Dachary Chief Research Officer // eNovance labs http://labs.enovance.com // ✉ l...@enovance.com ☎ +33 1 49 70 99 82 ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp
Re: [Openstack] [metering] delta vs. cumulative meters
On Tue, May 29, 2012 at 1:10 PM, Matt Joyce matt.jo...@cloudscaling.comwrote: Do we want a history of deltas or just last delta and leave the history to UI implementers? The collector should record everything and we should make the query API work so the caller can get what they want out of the datastore. On Tue, May 29, 2012 at 9:30 AM, Loic Dachary l...@enovance.com wrote: On 05/29/2012 05:42 PM, Doug Hellmann wrote: IIRC, the meters discussed in the wiki [1] are supposed to show delta values (usage since the last time an event was generated), although the Alternate Gauge Design section discusses cumulative meters instead. The libvirt pollsters we have now produce cumulative data, and it might be complicated and error-prone to change them to compute the deltas internally before generating events. On the other hand, some of the other counters may be more complicated to generate as cumulative values. Rather than taking an either/or approach, I propose we document for each counter whether it uses the delta or cumulative approach. The API server can ask the meter plugin for that piece of information when calculating the aggregate value so that the caller does not have to keep track of the rules for each meter. I tend to prefer using a gauge and I agree that documenting what was preferred (gauge or delta) for a given meter / counter is enough. My 2ct ;-) -- Loïc Dachary Chief Research Officer // eNovance labs http://labs.enovance.com // ✉ l...@enovance.com ☎ +33 1 49 70 99 82 ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp
Re: [Openstack] [metering] delta vs. cumulative meters
On 05/29/2012 07:58 PM, Doug Hellmann wrote: On Tue, May 29, 2012 at 12:30 PM, Loic Dachary l...@enovance.com mailto:l...@enovance.com wrote: On 05/29/2012 05:42 PM, Doug Hellmann wrote: IIRC, the meters discussed in the wiki [1] are supposed to show delta values (usage since the last time an event was generated), although the Alternate Gauge Design section discusses cumulative meters instead. The libvirt pollsters we have now produce cumulative data, and it might be complicated and error-prone to change them to compute the deltas internally before generating events. On the other hand, some of the other counters may be more complicated to generate as cumulative values. Rather than taking an either/or approach, I propose we document for each counter whether it uses the delta or cumulative approach. The API server can ask the meter plugin for that piece of information when calculating the aggregate value so that the caller does not have to keep track of the rules for each meter. I tend to prefer using a gauge and I agree that documenting what was preferred (gauge or delta) for a given meter / counter is enough. Just to be clear, by gauge do you mean the cumulative form of the counter? Yes, I do ;-) ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp
