Re: [openstack-dev] [nova] Dynamic scheduling
For those who are interested in dynamic scheduling, you can go to IBM booth today (5.12) from 10:45 to 12:45, we will have a simple demo there. There will also be a design session Future of Gantt APIs and interfaces which include run time policy for OpenStack at Friday, May 16 10:50am - 11:30am ( http://junodesignsummit.sched.org/event/36c19ae807a02ef7015ab042fd4541e6#.U3DNpChpdeM ) The google doc related to run time policy is here: https://docs.google.com/document/d/1DMsnGxQ3P-OwZCF3uxaUeEFaKX8LqUqmmgQ_7EVK7Y8/edit?usp=sharing Thanks! 2014-04-12 1:00 GMT+08:00 Tim Bell tim.b...@cern.ch: -Original Message- From: Andrew Laski [mailto:andrew.la...@rackspace.com] Sent: 11 April 2014 16:38 To: OpenStack Development Mailing List (not for usage questions) Subject: Re: [openstack-dev] [nova] Dynamic scheduling On 04/10/14 at 11:33pm, Oleg Gelbukh wrote: Andrew, ... In my opinion what's being proposed doesn't seem to fit cleanly into any existing service, so perhaps it could start as a standalone entity. Then once there's something that can be used and demoed a proper place might suggest itself, or it might make sense to keep it separate. I strongly support no auto scaling. Heat looks after this and is a user facing activity since it needs to know what to do when a new VM is created and how to set it up. A dynamic scheduling 'service' would work on an operator infrastructure layer performing VM relocation according to the service provider needs (balance between optimisation, thrashing, acceptable downtime). It should be performed within the SLA expectations of the VM. The dynamic scheduling is 'OpenStack Tetris', trying to ensure a consistent packing policy of VMs on resources based on the policy for the service class. Tim ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev -- Thanks, Jay ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [nova] Dynamic scheduling
On 04/10/14 at 11:33pm, Oleg Gelbukh wrote: Andrew, Thank you for clarification! On Thu, Apr 10, 2014 at 3:47 PM, Andrew Laski andrew.la...@rackspace.comwrote: The scheduler as it currently exists is a placement engine. There is sufficient complexity in the scheduler with just that responsibility so I would prefer to see anything that's making runtime decisions separated out. Perhaps it could just be another service within the scheduler project once it's broken out, but I think it will be beneficial to have a clear distinction between placement decisions and runtime monitoring. Do you think that auto-scaling could be considered another facet of this 'runtime monitoring' functionality? Now it is a combination of Heat and Ceilometer. Does it worth moving to hypothetical runtime mobility service as well? Auto-scaling is certainly a facet of runtime monitoring. But auto-scaling performs actions based on a set of user defined rules and is very visible while the enhancements proposed below are intended to benefit deployers and be very invisible to users. So the set of allowable actions is very constrained compared to what auto-scaling can do. In my opinion what's being proposed doesn't seem to fit cleanly into any existing service, so perhaps it could start as a standalone entity. Then once there's something that can be used and demoed a proper place might suggest itself, or it might make sense to keep it separate. -- Best regards, Oleg Gelbukh -- Best regards, Oleg Gelbukh On Wed, Apr 9, 2014 at 7:47 PM, Jay Lau jay.lau@gmail.com wrote: @Oleg, Till now, I'm not sure the target of Gantt, is it for initial placement policy or run time policy or both, can you help clarify? @Henrique, not sure if you know IBM PRS (Platform Resource Scheduler) [1], we have finished the dynamic scheduler in our Icehouse version (PRS 2.2), it has exactly the same feature as your described, we are planning a live demo for this feature in Atlanta Summit. I'm also writing some document for run time policy which will cover more run time policies for OpenStack, but not finished yet. (My shame for the slow progress). The related blueprint is [2], you can also get some discussion from [3] [1] http://www-01.ibm.com/common/ssi/cgi-bin/ssialias?infotype= ANsubtype=CAhtmlfid=897/ENUS213-590appname=USN [2] https://blueprints.launchpad.net/nova/+spec/resource- optimization-service [3] http://markmail.org/~jaylau/OpenStack-DRS Thanks. 2014-04-09 23:21 GMT+08:00 Oleg Gelbukh ogelb...@mirantis.com: Henrique, You should check out Gantt project [1], it could be exactly the place to implement such features. It is a generic cross-project Scheduler as a Service forked from Nova recently. [1] https://github.com/openstack/gantt -- Best regards, Oleg Gelbukh Mirantis Labs On Wed, Apr 9, 2014 at 6:41 PM, Henrique Truta henriquecostatr...@gmail.com wrote: Hello, everyone! I am currently a graduate student and member of a group of contributors to OpenStack. We believe that a dynamic scheduler could improve the efficiency of an OpenStack cloud, either by rebalancing nodes to maximize performance or to minimize the number of active hosts, in order to minimize energy costs. Therefore, we would like to propose a dynamic scheduling mechanism to Nova. The main idea is using the Ceilometer information (e.g. RAM, CPU, disk usage) through the ceilometer-client and dinamically decide whether a instance should be live migrated. This might me done as a Nova periodic task, which will be executed every once in a given period or as a new independent project. In both cases, the current Nova scheduler will not be affected, since this new scheduler will be pluggable. We have done a search and found no such initiative in the OpenStack BPs. Outside the community, we found only a recent IBM announcement for a similiar feature in one of its cloud products. A possible flow is: In the new scheduler, we periodically make a call to Nova, get the instance list from a specific host and, for each instance, we make a call to the ceilometer-client (e.g. $ ceilometer statistics -m cpu_util -q resource=$INSTANCE_ID) and then, according to some specific parameters configured by the user, analyze the meters and do the proper migrations. Do you have any comments or suggestions? -- Ítalo Henrique Costa Truta ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev -- Thanks, Jay ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing
Re: [openstack-dev] [nova] Dynamic scheduling
Is there anyone currently working on Neat/Gantt projects? I'd like to contribute to them, as well. 2014-04-11 11:37 GMT-03:00 Andrew Laski andrew.la...@rackspace.com: On 04/10/14 at 11:33pm, Oleg Gelbukh wrote: Andrew, Thank you for clarification! On Thu, Apr 10, 2014 at 3:47 PM, Andrew Laski andrew.la...@rackspace.com wrote: The scheduler as it currently exists is a placement engine. There is sufficient complexity in the scheduler with just that responsibility so I would prefer to see anything that's making runtime decisions separated out. Perhaps it could just be another service within the scheduler project once it's broken out, but I think it will be beneficial to have a clear distinction between placement decisions and runtime monitoring. Do you think that auto-scaling could be considered another facet of this 'runtime monitoring' functionality? Now it is a combination of Heat and Ceilometer. Does it worth moving to hypothetical runtime mobility service as well? Auto-scaling is certainly a facet of runtime monitoring. But auto-scaling performs actions based on a set of user defined rules and is very visible while the enhancements proposed below are intended to benefit deployers and be very invisible to users. So the set of allowable actions is very constrained compared to what auto-scaling can do. In my opinion what's being proposed doesn't seem to fit cleanly into any existing service, so perhaps it could start as a standalone entity. Then once there's something that can be used and demoed a proper place might suggest itself, or it might make sense to keep it separate. -- Best regards, Oleg Gelbukh -- Best regards, Oleg Gelbukh On Wed, Apr 9, 2014 at 7:47 PM, Jay Lau jay.lau@gmail.com wrote: @Oleg, Till now, I'm not sure the target of Gantt, is it for initial placement policy or run time policy or both, can you help clarify? @Henrique, not sure if you know IBM PRS (Platform Resource Scheduler) [1], we have finished the dynamic scheduler in our Icehouse version (PRS 2.2), it has exactly the same feature as your described, we are planning a live demo for this feature in Atlanta Summit. I'm also writing some document for run time policy which will cover more run time policies for OpenStack, but not finished yet. (My shame for the slow progress). The related blueprint is [2], you can also get some discussion from [3] [1] http://www-01.ibm.com/common/ssi/cgi-bin/ssialias?infotype= ANsubtype=CAhtmlfid=897/ENUS213-590appname=USN [2] https://blueprints.launchpad.net/nova/+spec/resource- optimization-service [3] http://markmail.org/~jaylau/OpenStack-DRS Thanks. 2014-04-09 23:21 GMT+08:00 Oleg Gelbukh ogelb...@mirantis.com: Henrique, You should check out Gantt project [1], it could be exactly the place to implement such features. It is a generic cross-project Scheduler as a Service forked from Nova recently. [1] https://github.com/openstack/gantt -- Best regards, Oleg Gelbukh Mirantis Labs On Wed, Apr 9, 2014 at 6:41 PM, Henrique Truta henriquecostatr...@gmail.com wrote: Hello, everyone! I am currently a graduate student and member of a group of contributors to OpenStack. We believe that a dynamic scheduler could improve the efficiency of an OpenStack cloud, either by rebalancing nodes to maximize performance or to minimize the number of active hosts, in order to minimize energy costs. Therefore, we would like to propose a dynamic scheduling mechanism to Nova. The main idea is using the Ceilometer information (e.g. RAM, CPU, disk usage) through the ceilometer-client and dinamically decide whether a instance should be live migrated. This might me done as a Nova periodic task, which will be executed every once in a given period or as a new independent project. In both cases, the current Nova scheduler will not be affected, since this new scheduler will be pluggable. We have done a search and found no such initiative in the OpenStack BPs. Outside the community, we found only a recent IBM announcement for a similiar feature in one of its cloud products. A possible flow is: In the new scheduler, we periodically make a call to Nova, get the instance list from a specific host and, for each instance, we make a call to the ceilometer-client (e.g. $ ceilometer statistics -m cpu_util -q resource=$INSTANCE_ID) and then, according to some specific parameters configured by the user, analyze the meters and do the proper migrations. Do you have any comments or suggestions? -- Ítalo Henrique Costa Truta ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [nova] Dynamic scheduling
-Original Message- From: Andrew Laski [mailto:andrew.la...@rackspace.com] Sent: 11 April 2014 16:38 To: OpenStack Development Mailing List (not for usage questions) Subject: Re: [openstack-dev] [nova] Dynamic scheduling On 04/10/14 at 11:33pm, Oleg Gelbukh wrote: Andrew, ... In my opinion what's being proposed doesn't seem to fit cleanly into any existing service, so perhaps it could start as a standalone entity. Then once there's something that can be used and demoed a proper place might suggest itself, or it might make sense to keep it separate. I strongly support no auto scaling. Heat looks after this and is a user facing activity since it needs to know what to do when a new VM is created and how to set it up. A dynamic scheduling 'service' would work on an operator infrastructure layer performing VM relocation according to the service provider needs (balance between optimisation, thrashing, acceptable downtime). It should be performed within the SLA expectations of the VM. The dynamic scheduling is 'OpenStack Tetris', trying to ensure a consistent packing policy of VMs on resources based on the policy for the service class. Tim ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [nova] Dynamic scheduling
Ditto, I am also interested in that area. We're implementing a framework to monitor different metrics from Ceilometer, apply predefined policies from administrators, and take actions if some conditions are met for resource optimization purpose or SLA purpose. Jay, is IBM PRS open source? Thanks. -- Shane From: Susanne Balle [mailto:sleipnir...@gmail.com] Sent: Thursday, April 10, 2014 12:57 AM To: OpenStack Development Mailing List (not for usage questions) Subject: Re: [openstack-dev] [nova] Dynamic scheduling Ditto. I am interested in contributing as well. Does Gant work with Devstack? I am assuming the link will give me directions on how to test it and contribute to the project. Susanne On Wed, Apr 9, 2014 at 12:44 PM, Henrique Truta henriquecostatr...@gmail.commailto:henriquecostatr...@gmail.com wrote: @Oleg, @Sylvain, @Leandro, Thanls. I'll check the Gantt project and the blueprint 2014-04-09 12:59 GMT-03:00 Sylvain Bauza sylvain.ba...@gmail.commailto:sylvain.ba...@gmail.com: 2014-04-09 17:47 GMT+02:00 Jay Lau jay.lau@gmail.commailto:jay.lau@gmail.com: @Oleg, Till now, I'm not sure the target of Gantt, is it for initial placement policy or run time policy or both, can you help clarify? I don't want to talk on behalf of Oleg, but Gantt is targeted to be the forklift of the current Nova scheduler. So, a placement decision based on dynamic metrics would be worth it. That said, as Gantt is not targeted to be delivered until Juno at least (with Nova sched deprecated), I think any progress on a BP should target Nova with respect to the forklift efforts, so it would automatically be ported to Gantt once the actual fork would happen. -Sylvain Jay ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.orgmailto:OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.orgmailto:OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev -- -- Ítalo Henrique Costa Truta ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.orgmailto:OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [nova] Dynamic scheduling
Hello, Jay, As a fork of nova-scheduler, Gantt most likely will handle initial placement. However, even nova-scheduler now supports some runtime operations (for example, scheduling of evacuated/migrated instances). Given the runtime scheduling arises in this list regularly, I guess such features will make their way into Scheduler service eventually. -- Best regards, Oleg Gelbukh On Wed, Apr 9, 2014 at 7:47 PM, Jay Lau jay.lau@gmail.com wrote: @Oleg, Till now, I'm not sure the target of Gantt, is it for initial placement policy or run time policy or both, can you help clarify? @Henrique, not sure if you know IBM PRS (Platform Resource Scheduler) [1], we have finished the dynamic scheduler in our Icehouse version (PRS 2.2), it has exactly the same feature as your described, we are planning a live demo for this feature in Atlanta Summit. I'm also writing some document for run time policy which will cover more run time policies for OpenStack, but not finished yet. (My shame for the slow progress). The related blueprint is [2], you can also get some discussion from [3] [1] http://www-01.ibm.com/common/ssi/cgi-bin/ssialias?infotype=ANsubtype=CAhtmlfid=897/ENUS213-590appname=USN [2] https://blueprints.launchpad.net/nova/+spec/resource-optimization-service [3] http://markmail.org/~jaylau/OpenStack-DRS Thanks. 2014-04-09 23:21 GMT+08:00 Oleg Gelbukh ogelb...@mirantis.com: Henrique, You should check out Gantt project [1], it could be exactly the place to implement such features. It is a generic cross-project Scheduler as a Service forked from Nova recently. [1] https://github.com/openstack/gantt -- Best regards, Oleg Gelbukh Mirantis Labs On Wed, Apr 9, 2014 at 6:41 PM, Henrique Truta henriquecostatr...@gmail.com wrote: Hello, everyone! I am currently a graduate student and member of a group of contributors to OpenStack. We believe that a dynamic scheduler could improve the efficiency of an OpenStack cloud, either by rebalancing nodes to maximize performance or to minimize the number of active hosts, in order to minimize energy costs. Therefore, we would like to propose a dynamic scheduling mechanism to Nova. The main idea is using the Ceilometer information (e.g. RAM, CPU, disk usage) through the ceilometer-client and dinamically decide whether a instance should be live migrated. This might me done as a Nova periodic task, which will be executed every once in a given period or as a new independent project. In both cases, the current Nova scheduler will not be affected, since this new scheduler will be pluggable. We have done a search and found no such initiative in the OpenStack BPs. Outside the community, we found only a recent IBM announcement for a similiar feature in one of its cloud products. A possible flow is: In the new scheduler, we periodically make a call to Nova, get the instance list from a specific host and, for each instance, we make a call to the ceilometer-client (e.g. $ ceilometer statistics -m cpu_util -q resource=$INSTANCE_ID) and then, according to some specific parameters configured by the user, analyze the meters and do the proper migrations. Do you have any comments or suggestions? -- Ítalo Henrique Costa Truta ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev -- Thanks, Jay ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [nova] Dynamic scheduling
Is it a same thing like openstack-neat project? http://openstack-neat.org/ I am curious about why Neat was not accepted previously. -Hao From: Oleg Gelbukh [mailto:ogelb...@mirantis.com] Sent: Thursday, April 10, 2014 3:48 PM To: OpenStack Development Mailing List (not for usage questions) Subject: Re: [openstack-dev] [nova] Dynamic scheduling Hello, Jay, As a fork of nova-scheduler, Gantt most likely will handle initial placement. However, even nova-scheduler now supports some runtime operations (for example, scheduling of evacuated/migrated instances). Given the runtime scheduling arises in this list regularly, I guess such features will make their way into Scheduler service eventually. -- Best regards, Oleg Gelbukh On Wed, Apr 9, 2014 at 7:47 PM, Jay Lau jay.lau@gmail.commailto:jay.lau@gmail.com wrote: @Oleg, Till now, I'm not sure the target of Gantt, is it for initial placement policy or run time policy or both, can you help clarify? @Henrique, not sure if you know IBM PRS (Platform Resource Scheduler) [1], we have finished the dynamic scheduler in our Icehouse version (PRS 2.2), it has exactly the same feature as your described, we are planning a live demo for this feature in Atlanta Summit. I'm also writing some document for run time policy which will cover more run time policies for OpenStack, but not finished yet. (My shame for the slow progress). The related blueprint is [2], you can also get some discussion from [3] [1] http://www-01.ibm.com/common/ssi/cgi-bin/ssialias?infotype=ANsubtype=CAhtmlfid=897/ENUS213-590appname=USN [2] https://blueprints.launchpad.net/nova/+spec/resource-optimization-service [3] http://markmail.org/~jaylau/OpenStack-DRS Thanks. 2014-04-09 23:21 GMT+08:00 Oleg Gelbukh ogelb...@mirantis.commailto:ogelb...@mirantis.com: Henrique, You should check out Gantt project [1], it could be exactly the place to implement such features. It is a generic cross-project Scheduler as a Service forked from Nova recently. [1] https://github.com/openstack/gantt -- Best regards, Oleg Gelbukh Mirantis Labs On Wed, Apr 9, 2014 at 6:41 PM, Henrique Truta henriquecostatr...@gmail.commailto:henriquecostatr...@gmail.com wrote: Hello, everyone! I am currently a graduate student and member of a group of contributors to OpenStack. We believe that a dynamic scheduler could improve the efficiency of an OpenStack cloud, either by rebalancing nodes to maximize performance or to minimize the number of active hosts, in order to minimize energy costs. Therefore, we would like to propose a dynamic scheduling mechanism to Nova. The main idea is using the Ceilometer information (e.g. RAM, CPU, disk usage) through the ceilometer-client and dinamically decide whether a instance should be live migrated. This might me done as a Nova periodic task, which will be executed every once in a given period or as a new independent project. In both cases, the current Nova scheduler will not be affected, since this new scheduler will be pluggable. We have done a search and found no such initiative in the OpenStack BPs. Outside the community, we found only a recent IBM announcement for a similiar feature in one of its cloud products. A possible flow is: In the new scheduler, we periodically make a call to Nova, get the instance list from a specific host and, for each instance, we make a call to the ceilometer-client (e.g. $ ceilometer statistics -m cpu_util -q resource=$INSTANCE_ID) and then, according to some specific parameters configured by the user, analyze the meters and do the proper migrations. Do you have any comments or suggestions? -- Ítalo Henrique Costa Truta ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.orgmailto:OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.orgmailto:OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev -- Thanks, Jay ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.orgmailto:OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [nova] Dynamic scheduling
Is the [1] mentioned below open source? Interested in it :) Also, my own opinion is nova is a cloud solution that aim to control virtual / real machine lifecycle management the dynamic scheduling mechanism is something like optimization of the cloud resource, so maybe incubate in other places is better Best Regards! Kevin (Chen) Ji 纪 晨 Engineer, zVM Development, CSTL Notes: Chen CH Ji/China/IBM@IBMCN Internet: jiche...@cn.ibm.com Phone: +86-10-82454158 Address: 3/F Ring Building, ZhongGuanCun Software Park, Haidian District, Beijing 100193, PRC From: Jay Lau jay.lau@gmail.com To: OpenStack Development Mailing List (not for usage questions) openstack-dev@lists.openstack.org, Date: 04/09/2014 11:45 PM Subject:Re: [openstack-dev] [nova] Dynamic scheduling @Oleg, Till now, I'm not sure the target of Gantt, is it for initial placement policy or run time policy or both, can you help clarify? @Henrique, not sure if you know IBM PRS (Platform Resource Scheduler) [1], we have finished the dynamic scheduler in our Icehouse version (PRS 2.2), it has exactly the same feature as your described, we are planning a live demo for this feature in Atlanta Summit. I'm also writing some document for run time policy which will cover more run time policies for OpenStack, but not finished yet. (My shame for the slow progress). The related blueprint is [2], you can also get some discussion from [3] [1] http://www-01.ibm.com/common/ssi/cgi-bin/ssialias?infotype=ANsubtype=CAhtmlfid=897/ENUS213-590appname=USN [2] https://blueprints.launchpad.net/nova/+spec/resource-optimization-service [3] http://markmail.org/~jaylau/OpenStack-DRS Thanks. 2014-04-09 23:21 GMT+08:00 Oleg Gelbukh ogelb...@mirantis.com: Henrique, You should check out Gantt project [1], it could be exactly the place to implement such features. It is a generic cross-project Scheduler as a Service forked from Nova recently. [1] https://github.com/openstack/gantt -- Best regards, Oleg Gelbukh Mirantis Labs On Wed, Apr 9, 2014 at 6:41 PM, Henrique Truta henriquecostatr...@gmail.com wrote: Hello, everyone! I am currently a graduate student and member of a group of contributors to OpenStack. We believe that a dynamic scheduler could improve the efficiency of an OpenStack cloud, either by rebalancing nodes to maximize performance or to minimize the number of active hosts, in order to minimize energy costs. Therefore, we would like to propose a dynamic scheduling mechanism to Nova. The main idea is using the Ceilometer information (e.g. RAM, CPU, disk usage) through the ceilometer-client and dinamically decide whether a instance should be live migrated. This might me done as a Nova periodic task, which will be executed every once in a given period or as a new independent project. In both cases, the current Nova scheduler will not be affected, since this new scheduler will be pluggable. We have done a search and found no such initiative in the OpenStack BPs. Outside the community, we found only a recent IBM announcement for a similiar feature in one of its cloud products. A possible flow is: In the new scheduler, we periodically make a call to Nova, get the instance list from a specific host and, for each instance, we make a call to the ceilometer-client (e.g. $ ceilometer statistics -m cpu_util -q resource=$INSTANCE_ID) and then, according to some specific parameters configured by the user, analyze the meters and do the proper migrations. Do you have any comments or suggestions? -- Ítalo Henrique Costa Truta ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev -- Thanks, Jay___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev inline: graycol.gif___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [nova] Dynamic scheduling
+1 A reference case is that, in an operating system like Linux, hardware drivers only abstract hardware devices into manageable resources with standard interfaces. Then the kernel takes the job to allocate resources, map and schedule workload onto them, and so on. In OpenStack, projects for resource abstraction and virtualization, including Nova, Cinder, Neutron, should be considered more like Linux device drivers, instead of the Linux kernel itself. Therefore, the scheduling functions and services could be separated, which is helpful from the long-term evaluation point of view. Best Regards, Yu Zhang Huawei Technologies Co., Ltd. Phone: 86-10-82829573 Email: zhangy...@huawei.com http://www.huawei.com From: Chen CH Ji [mailto:jiche...@cn.ibm.com] Sent: Thursday, April 10, 2014 5:08 PM To: OpenStack Development Mailing List (not for usage questions) Subject: Re: [openstack-dev] [nova] Dynamic scheduling Is the [1] mentioned below open source? Interested in it :) Also, my own opinion is nova is a cloud solution that aim to control virtual / real machine lifecycle management the dynamic scheduling mechanism is something like optimization of the cloud resource, so maybe incubate in other places is better Best Regards! Kevin (Chen) Ji 纪 晨 Engineer, zVM Development, CSTL Notes: Chen CH Ji/China/IBM@IBMCN Internet: jiche...@cn.ibm.commailto:jiche...@cn.ibm.com Phone: +86-10-82454158 Address: 3/F Ring Building, ZhongGuanCun Software Park, Haidian District, Beijing 100193, PRC [Inactive hide details for Jay Lau ---04/09/2014 11:45:12 PM---@Oleg, Till now, I'm not sure the target of Gantt, is it for init]Jay Lau ---04/09/2014 11:45:12 PM---@Oleg, Till now, I'm not sure the target of Gantt, is it for initial placement policy or run time po From: Jay Lau jay.lau@gmail.commailto:jay.lau@gmail.com To: OpenStack Development Mailing List (not for usage questions) openstack-dev@lists.openstack.orgmailto:openstack-dev@lists.openstack.org, Date: 04/09/2014 11:45 PM Subject: Re: [openstack-dev] [nova] Dynamic scheduling @Oleg, Till now, I'm not sure the target of Gantt, is it for initial placement policy or run time policy or both, can you help clarify? @Henrique, not sure if you know IBM PRS (Platform Resource Scheduler) [1], we have finished the dynamic scheduler in our Icehouse version (PRS 2.2), it has exactly the same feature as your described, we are planning a live demo for this feature in Atlanta Summit. I'm also writing some document for run time policy which will cover more run time policies for OpenStack, but not finished yet. (My shame for the slow progress). The related blueprint is [2], you can also get some discussion from [3] [1] http://www-01.ibm.com/common/ssi/cgi-bin/ssialias?infotype=ANsubtype=CAhtmlfid=897/ENUS213-590appname=USN [2] https://blueprints.launchpad.net/nova/+spec/resource-optimization-service [3] http://markmail.org/~jaylau/OpenStack-DRS Thanks. 2014-04-09 23:21 GMT+08:00 Oleg Gelbukh ogelb...@mirantis.commailto:ogelb...@mirantis.com: Henrique, You should check out Gantt project [1], it could be exactly the place to implement such features. It is a generic cross-project Scheduler as a Service forked from Nova recently. [1] https://github.com/openstack/gantt -- Best regards, Oleg Gelbukh Mirantis Labs On Wed, Apr 9, 2014 at 6:41 PM, Henrique Truta henriquecostatr...@gmail.commailto:henriquecostatr...@gmail.com wrote: Hello, everyone! I am currently a graduate student and member of a group of contributors to OpenStack. We believe that a dynamic scheduler could improve the efficiency of an OpenStack cloud, either by rebalancing nodes to maximize performance or to minimize the number of active hosts, in order to minimize energy costs. Therefore, we would like to propose a dynamic scheduling mechanism to Nova. The main idea is using the Ceilometer information (e.g. RAM, CPU, disk usage) through the ceilometer-client and dinamically decide whether a instance should be live migrated. This might me done as a Nova periodic task, which will be executed every once in a given period or as a new independent project. In both cases, the current Nova scheduler will not be affected, since this new scheduler will be pluggable. We have done a search and found no such initiative in the OpenStack BPs. Outside the community, we found only a recent IBM announcement for a similiar feature in one of its cloud products. A possible flow is: In the new scheduler, we periodically make a call to Nova, get the instance list from a specific host and, for each instance, we make a call to the ceilometer-client (e.g. $ ceilometer statistics -m cpu_util -q resource=$INSTANCE_ID) and then, according to some specific parameters configured by the user, analyze the meters and do the proper migrations. Do you have any comments or suggestions? -- Ítalo Henrique Costa Truta
Re: [openstack-dev] [nova] Dynamic scheduling
On 04/10/14 at 11:48am, Oleg Gelbukh wrote: Hello, Jay, As a fork of nova-scheduler, Gantt most likely will handle initial placement. However, even nova-scheduler now supports some runtime operations (for example, scheduling of evacuated/migrated instances). Please note that the operations you've listed are all user initiated actions, and therefore differ from what's being proposed. There is currently nothing in Nova that performs an action on an ACTIVE VM without user input. Given the runtime scheduling arises in this list regularly, I guess such features will make their way into Scheduler service eventually. The scheduler as it currently exists is a placement engine. There is sufficient complexity in the scheduler with just that responsibility so I would prefer to see anything that's making runtime decisions separated out. Perhaps it could just be another service within the scheduler project once it's broken out, but I think it will be beneficial to have a clear distinction between placement decisions and runtime monitoring. -- Best regards, Oleg Gelbukh On Wed, Apr 9, 2014 at 7:47 PM, Jay Lau jay.lau@gmail.com wrote: @Oleg, Till now, I'm not sure the target of Gantt, is it for initial placement policy or run time policy or both, can you help clarify? @Henrique, not sure if you know IBM PRS (Platform Resource Scheduler) [1], we have finished the dynamic scheduler in our Icehouse version (PRS 2.2), it has exactly the same feature as your described, we are planning a live demo for this feature in Atlanta Summit. I'm also writing some document for run time policy which will cover more run time policies for OpenStack, but not finished yet. (My shame for the slow progress). The related blueprint is [2], you can also get some discussion from [3] [1] http://www-01.ibm.com/common/ssi/cgi-bin/ssialias?infotype=ANsubtype=CAhtmlfid=897/ENUS213-590appname=USN [2] https://blueprints.launchpad.net/nova/+spec/resource-optimization-service [3] http://markmail.org/~jaylau/OpenStack-DRS Thanks. 2014-04-09 23:21 GMT+08:00 Oleg Gelbukh ogelb...@mirantis.com: Henrique, You should check out Gantt project [1], it could be exactly the place to implement such features. It is a generic cross-project Scheduler as a Service forked from Nova recently. [1] https://github.com/openstack/gantt -- Best regards, Oleg Gelbukh Mirantis Labs On Wed, Apr 9, 2014 at 6:41 PM, Henrique Truta henriquecostatr...@gmail.com wrote: Hello, everyone! I am currently a graduate student and member of a group of contributors to OpenStack. We believe that a dynamic scheduler could improve the efficiency of an OpenStack cloud, either by rebalancing nodes to maximize performance or to minimize the number of active hosts, in order to minimize energy costs. Therefore, we would like to propose a dynamic scheduling mechanism to Nova. The main idea is using the Ceilometer information (e.g. RAM, CPU, disk usage) through the ceilometer-client and dinamically decide whether a instance should be live migrated. This might me done as a Nova periodic task, which will be executed every once in a given period or as a new independent project. In both cases, the current Nova scheduler will not be affected, since this new scheduler will be pluggable. We have done a search and found no such initiative in the OpenStack BPs. Outside the community, we found only a recent IBM announcement for a similiar feature in one of its cloud products. A possible flow is: In the new scheduler, we periodically make a call to Nova, get the instance list from a specific host and, for each instance, we make a call to the ceilometer-client (e.g. $ ceilometer statistics -m cpu_util -q resource=$INSTANCE_ID) and then, according to some specific parameters configured by the user, analyze the meters and do the proper migrations. Do you have any comments or suggestions? -- Ítalo Henrique Costa Truta ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev -- Thanks, Jay ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [nova] Dynamic scheduling
2014-04-10 0:57 GMT+08:00 Susanne Balle sleipnir...@gmail.com: Ditto. I am interested in contributing as well. Does Gant work with Devstack? I am assuming the link will give me directions on how to test it and contribute to the project. You can refer to http://lists.openstack.org/pipermail/openstack-dev/2014-February/026330.htmlfor how to enable Gantt with devstack. Susanne On Wed, Apr 9, 2014 at 12:44 PM, Henrique Truta henriquecostatr...@gmail.com wrote: @Oleg, @Sylvain, @Leandro, Thanls. I'll check the Gantt project and the blueprint 2014-04-09 12:59 GMT-03:00 Sylvain Bauza sylvain.ba...@gmail.com: 2014-04-09 17:47 GMT+02:00 Jay Lau jay.lau@gmail.com: @Oleg, Till now, I'm not sure the target of Gantt, is it for initial placement policy or run time policy or both, can you help clarify? I don't want to talk on behalf of Oleg, but Gantt is targeted to be the forklift of the current Nova scheduler. So, a placement decision based on dynamic metrics would be worth it. That said, as Gantt is not targeted to be delivered until Juno at least (with Nova sched deprecated), I think any progress on a BP should target Nova with respect to the forklift efforts, so it would automatically be ported to Gantt once the actual fork would happen. -Sylvain Jay ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev -- -- Ítalo Henrique Costa Truta ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev -- Thanks, Jay ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [nova] Dynamic scheduling
Hi Shane, IBM PRS is not open source, its an add on product for OpenStack. For your framework, ceilometer can only report some VM metrics, how did you do the resource optimization based on only vm metrics? Thanks. 2014-04-10 14:57 GMT+08:00 Wang, Shane shane.w...@intel.com: Ditto, I am also interested in that area. We're implementing a framework to monitor different metrics from Ceilometer, apply predefined policies from administrators, and take actions if some conditions are met for resource optimization purpose or SLA purpose. Jay, is IBM PRS open source? Thanks. -- Shane *From:* Susanne Balle [mailto:sleipnir...@gmail.com] *Sent:* Thursday, April 10, 2014 12:57 AM *To:* OpenStack Development Mailing List (not for usage questions) *Subject:* Re: [openstack-dev] [nova] Dynamic scheduling Ditto. I am interested in contributing as well. Does Gant work with Devstack? I am assuming the link will give me directions on how to test it and contribute to the project. Susanne On Wed, Apr 9, 2014 at 12:44 PM, Henrique Truta henriquecostatr...@gmail.com wrote: @Oleg, @Sylvain, @Leandro, Thanls. I'll check the Gantt project and the blueprint 2014-04-09 12:59 GMT-03:00 Sylvain Bauza sylvain.ba...@gmail.com: 2014-04-09 17:47 GMT+02:00 Jay Lau jay.lau@gmail.com: @Oleg, Till now, I'm not sure the target of Gantt, is it for initial placement policy or run time policy or both, can you help clarify? I don't want to talk on behalf of Oleg, but Gantt is targeted to be the forklift of the current Nova scheduler. So, a placement decision based on dynamic metrics would be worth it. That said, as Gantt is not targeted to be delivered until Juno at least (with Nova sched deprecated), I think any progress on a BP should target Nova with respect to the forklift efforts, so it would automatically be ported to Gantt once the actual fork would happen. -Sylvain Jay ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev -- -- Ítalo Henrique Costa Truta ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev -- Thanks, Jay ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [nova] Dynamic scheduling
2014-04-10 16:16 GMT+02:00 Jay Lau jay.lau@gmail.com: 2014-04-10 0:57 GMT+08:00 Susanne Balle sleipnir...@gmail.com: Ditto. I am interested in contributing as well. Does Gant work with Devstack? I am assuming the link will give me directions on how to test it and contribute to the project. You can refer to http://lists.openstack.org/pipermail/openstack-dev/2014-February/026330.htmlfor how to enable Gantt with devstack. The Gantt code is currently freezed to a fork happened in November. As said above, IMHO there is no sense to use devstack for testing Gantt as the current repo is only a ghost one... -Sylvain ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [nova] Dynamic scheduling
Andrew, Thank you for clarification! On Thu, Apr 10, 2014 at 3:47 PM, Andrew Laski andrew.la...@rackspace.comwrote: The scheduler as it currently exists is a placement engine. There is sufficient complexity in the scheduler with just that responsibility so I would prefer to see anything that's making runtime decisions separated out. Perhaps it could just be another service within the scheduler project once it's broken out, but I think it will be beneficial to have a clear distinction between placement decisions and runtime monitoring. Do you think that auto-scaling could be considered another facet of this 'runtime monitoring' functionality? Now it is a combination of Heat and Ceilometer. Does it worth moving to hypothetical runtime mobility service as well? -- Best regards, Oleg Gelbukh -- Best regards, Oleg Gelbukh On Wed, Apr 9, 2014 at 7:47 PM, Jay Lau jay.lau@gmail.com wrote: @Oleg, Till now, I'm not sure the target of Gantt, is it for initial placement policy or run time policy or both, can you help clarify? @Henrique, not sure if you know IBM PRS (Platform Resource Scheduler) [1], we have finished the dynamic scheduler in our Icehouse version (PRS 2.2), it has exactly the same feature as your described, we are planning a live demo for this feature in Atlanta Summit. I'm also writing some document for run time policy which will cover more run time policies for OpenStack, but not finished yet. (My shame for the slow progress). The related blueprint is [2], you can also get some discussion from [3] [1] http://www-01.ibm.com/common/ssi/cgi-bin/ssialias?infotype= ANsubtype=CAhtmlfid=897/ENUS213-590appname=USN [2] https://blueprints.launchpad.net/nova/+spec/resource- optimization-service [3] http://markmail.org/~jaylau/OpenStack-DRS Thanks. 2014-04-09 23:21 GMT+08:00 Oleg Gelbukh ogelb...@mirantis.com: Henrique, You should check out Gantt project [1], it could be exactly the place to implement such features. It is a generic cross-project Scheduler as a Service forked from Nova recently. [1] https://github.com/openstack/gantt -- Best regards, Oleg Gelbukh Mirantis Labs On Wed, Apr 9, 2014 at 6:41 PM, Henrique Truta henriquecostatr...@gmail.com wrote: Hello, everyone! I am currently a graduate student and member of a group of contributors to OpenStack. We believe that a dynamic scheduler could improve the efficiency of an OpenStack cloud, either by rebalancing nodes to maximize performance or to minimize the number of active hosts, in order to minimize energy costs. Therefore, we would like to propose a dynamic scheduling mechanism to Nova. The main idea is using the Ceilometer information (e.g. RAM, CPU, disk usage) through the ceilometer-client and dinamically decide whether a instance should be live migrated. This might me done as a Nova periodic task, which will be executed every once in a given period or as a new independent project. In both cases, the current Nova scheduler will not be affected, since this new scheduler will be pluggable. We have done a search and found no such initiative in the OpenStack BPs. Outside the community, we found only a recent IBM announcement for a similiar feature in one of its cloud products. A possible flow is: In the new scheduler, we periodically make a call to Nova, get the instance list from a specific host and, for each instance, we make a call to the ceilometer-client (e.g. $ ceilometer statistics -m cpu_util -q resource=$INSTANCE_ID) and then, according to some specific parameters configured by the user, analyze the meters and do the proper migrations. Do you have any comments or suggestions? -- Ítalo Henrique Costa Truta ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev -- Thanks, Jay ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [nova] Dynamic scheduling
- Original Message - Is it a same thing like openstack-neat project? http://openstack-neat.org/ I am curious about why Neat was not accepted previously. -Hao Did anyone ever try to submit it? Looks useful though. -Steve ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [nova] Dynamic scheduling
- Original Message - - Original Message - Is it a same thing like openstack-neat project? http://openstack-neat.org/ I am curious about why Neat was not accepted previously. -Hao Did anyone ever try to submit it? Looks useful though. -Steve Nevermind, I probably found my answer: The source code is available under the Apache 2.0 license, but some methods implemented in the software are a part of a patent application. That means, commercial users need to obtain a license to use the IP noted in the patent. -Steve ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [nova] Dynamic scheduling
2014-04-11 3:54 GMT+08:00 Steve Gordon sgor...@redhat.com: - Original Message - Is it a same thing like openstack-neat project? http://openstack-neat.org/ I am curious about why Neat was not accepted previously. -Hao Did anyone ever try to submit it? Looks useful though. I submitted one cross project session which might already cover this: http://summit.openstack.org/cfp/details/262 -Steve ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev -- Thanks, Jay ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [nova] Dynamic scheduling
2014-04-11 0:22 GMT+08:00 Sylvain Bauza sylvain.ba...@gmail.com: 2014-04-10 16:16 GMT+02:00 Jay Lau jay.lau@gmail.com: 2014-04-10 0:57 GMT+08:00 Susanne Balle sleipnir...@gmail.com: Ditto. I am interested in contributing as well. Does Gant work with Devstack? I am assuming the link will give me directions on how to test it and contribute to the project. You can refer to http://lists.openstack.org/pipermail/openstack-dev/2014-February/026330.htmlfor how to enable Gantt with devstack. The Gantt code is currently freezed to a fork happened in November. As said above, IMHO there is no sense to use devstack for testing Gantt as the current repo is only a ghost one... OK, I see, thanks Sylvain. -Sylvain ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev -- Thanks, Jay ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [nova] Dynamic scheduling
Henrique, You should check out Gantt project [1], it could be exactly the place to implement such features. It is a generic cross-project Scheduler as a Service forked from Nova recently. [1] https://github.com/openstack/gantt -- Best regards, Oleg Gelbukh Mirantis Labs On Wed, Apr 9, 2014 at 6:41 PM, Henrique Truta henriquecostatr...@gmail.com wrote: Hello, everyone! I am currently a graduate student and member of a group of contributors to OpenStack. We believe that a dynamic scheduler could improve the efficiency of an OpenStack cloud, either by rebalancing nodes to maximize performance or to minimize the number of active hosts, in order to minimize energy costs. Therefore, we would like to propose a dynamic scheduling mechanism to Nova. The main idea is using the Ceilometer information (e.g. RAM, CPU, disk usage) through the ceilometer-client and dinamically decide whether a instance should be live migrated. This might me done as a Nova periodic task, which will be executed every once in a given period or as a new independent project. In both cases, the current Nova scheduler will not be affected, since this new scheduler will be pluggable. We have done a search and found no such initiative in the OpenStack BPs. Outside the community, we found only a recent IBM announcement for a similiar feature in one of its cloud products. A possible flow is: In the new scheduler, we periodically make a call to Nova, get the instance list from a specific host and, for each instance, we make a call to the ceilometer-client (e.g. $ ceilometer statistics -m cpu_util -q resource=$INSTANCE_ID) and then, according to some specific parameters configured by the user, analyze the meters and do the proper migrations. Do you have any comments or suggestions? -- Ítalo Henrique Costa Truta ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [nova] Dynamic scheduling
Hi Enrique, This bp [1] may cover the uses cases you are proposing. ( maybe not using ceilometer ) Also, you can take a look at openstack-neat [2] ( outside project ), that try to achieve something similar, but seems to be outdated. There's another initiate to have an external scheduler (GANTT) , so maybe there could be some place there for this kind of functionality. [1] https://blueprints.launchpad.net/nova/+spec/resource-optimization-service [2] http://openstack-neat.org El 09/04/2014 11:41 a.m., Henrique Truta escribió: Hello, everyone! I am currently a graduate student and member of a group of contributors to OpenStack. We believe that a dynamic scheduler could improve the efficiency of an OpenStack cloud, either by rebalancing nodes to maximize performance or to minimize the number of active hosts, in order to minimize energy costs. Therefore, we would like to propose a dynamic scheduling mechanism to Nova. The main idea is using the Ceilometer information (e.g. RAM, CPU, disk usage) through the ceilometer-client and dinamically decide whether a instance should be live migrated. This might me done as a Nova periodic task, which will be executed every once in a given period or as a new independent project. In both cases, the current Nova scheduler will not be affected, since this new scheduler will be pluggable. We have done a search and found no such initiative in the OpenStack BPs. Outside the community, we found only a recent IBM announcement for a similiar feature in one of its cloud products. A possible flow is: In the new scheduler, we periodically make a call to Nova, get the instance list from a specific host and, for each instance, we make a call to the ceilometer-client (e.g. $ ceilometer statistics -m cpu_util -q resource=$INSTANCE_ID) and then, according to some specific parameters configured by the user, analyze the meters and do the proper migrations. Do you have any comments or suggestions? -- Ítalo Henrique Costa Truta ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [nova] Dynamic scheduling
@Oleg, Till now, I'm not sure the target of Gantt, is it for initial placement policy or run time policy or both, can you help clarify? @Henrique, not sure if you know IBM PRS (Platform Resource Scheduler) [1], we have finished the dynamic scheduler in our Icehouse version (PRS 2.2), it has exactly the same feature as your described, we are planning a live demo for this feature in Atlanta Summit. I'm also writing some document for run time policy which will cover more run time policies for OpenStack, but not finished yet. (My shame for the slow progress). The related blueprint is [2], you can also get some discussion from [3] [1] http://www-01.ibm.com/common/ssi/cgi-bin/ssialias?infotype=ANsubtype=CAhtmlfid=897/ENUS213-590appname=USN [2] https://blueprints.launchpad.net/nova/+spec/resource-optimization-service [3] http://markmail.org/~jaylau/OpenStack-DRS Thanks. 2014-04-09 23:21 GMT+08:00 Oleg Gelbukh ogelb...@mirantis.com: Henrique, You should check out Gantt project [1], it could be exactly the place to implement such features. It is a generic cross-project Scheduler as a Service forked from Nova recently. [1] https://github.com/openstack/gantt -- Best regards, Oleg Gelbukh Mirantis Labs On Wed, Apr 9, 2014 at 6:41 PM, Henrique Truta henriquecostatr...@gmail.com wrote: Hello, everyone! I am currently a graduate student and member of a group of contributors to OpenStack. We believe that a dynamic scheduler could improve the efficiency of an OpenStack cloud, either by rebalancing nodes to maximize performance or to minimize the number of active hosts, in order to minimize energy costs. Therefore, we would like to propose a dynamic scheduling mechanism to Nova. The main idea is using the Ceilometer information (e.g. RAM, CPU, disk usage) through the ceilometer-client and dinamically decide whether a instance should be live migrated. This might me done as a Nova periodic task, which will be executed every once in a given period or as a new independent project. In both cases, the current Nova scheduler will not be affected, since this new scheduler will be pluggable. We have done a search and found no such initiative in the OpenStack BPs. Outside the community, we found only a recent IBM announcement for a similiar feature in one of its cloud products. A possible flow is: In the new scheduler, we periodically make a call to Nova, get the instance list from a specific host and, for each instance, we make a call to the ceilometer-client (e.g. $ ceilometer statistics -m cpu_util -q resource=$INSTANCE_ID) and then, according to some specific parameters configured by the user, analyze the meters and do the proper migrations. Do you have any comments or suggestions? -- Ítalo Henrique Costa Truta ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev -- Thanks, Jay ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [nova] Dynamic scheduling
2014-04-09 17:47 GMT+02:00 Jay Lau jay.lau@gmail.com: @Oleg, Till now, I'm not sure the target of Gantt, is it for initial placement policy or run time policy or both, can you help clarify? I don't want to talk on behalf of Oleg, but Gantt is targeted to be the forklift of the current Nova scheduler. So, a placement decision based on dynamic metrics would be worth it. That said, as Gantt is not targeted to be delivered until Juno at least (with Nova sched deprecated), I think any progress on a BP should target Nova with respect to the forklift efforts, so it would automatically be ported to Gantt once the actual fork would happen. -Sylvain Jay ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [nova] Dynamic scheduling
@Oleg, @Sylvain, @Leandro, Thanls. I'll check the Gantt project and the blueprint 2014-04-09 12:59 GMT-03:00 Sylvain Bauza sylvain.ba...@gmail.com: 2014-04-09 17:47 GMT+02:00 Jay Lau jay.lau@gmail.com: @Oleg, Till now, I'm not sure the target of Gantt, is it for initial placement policy or run time policy or both, can you help clarify? I don't want to talk on behalf of Oleg, but Gantt is targeted to be the forklift of the current Nova scheduler. So, a placement decision based on dynamic metrics would be worth it. That said, as Gantt is not targeted to be delivered until Juno at least (with Nova sched deprecated), I think any progress on a BP should target Nova with respect to the forklift efforts, so it would automatically be ported to Gantt once the actual fork would happen. -Sylvain Jay ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev -- -- Ítalo Henrique Costa Truta ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [nova] Dynamic scheduling
Ditto. I am interested in contributing as well. Does Gant work with Devstack? I am assuming the link will give me directions on how to test it and contribute to the project. Susanne On Wed, Apr 9, 2014 at 12:44 PM, Henrique Truta henriquecostatr...@gmail.com wrote: @Oleg, @Sylvain, @Leandro, Thanls. I'll check the Gantt project and the blueprint 2014-04-09 12:59 GMT-03:00 Sylvain Bauza sylvain.ba...@gmail.com: 2014-04-09 17:47 GMT+02:00 Jay Lau jay.lau@gmail.com: @Oleg, Till now, I'm not sure the target of Gantt, is it for initial placement policy or run time policy or both, can you help clarify? I don't want to talk on behalf of Oleg, but Gantt is targeted to be the forklift of the current Nova scheduler. So, a placement decision based on dynamic metrics would be worth it. That said, as Gantt is not targeted to be delivered until Juno at least (with Nova sched deprecated), I think any progress on a BP should target Nova with respect to the forklift efforts, so it would automatically be ported to Gantt once the actual fork would happen. -Sylvain Jay ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev -- -- Ítalo Henrique Costa Truta ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [nova] Dynamic scheduling
2014-04-09 18:57 GMT+02:00 Susanne Balle sleipnir...@gmail.com: Does Gant work with Devstack? I am assuming the link will give me directions on how to test it and contribute to the project. https://github.com/openstack/gantt/blob/master/README.rst#disclaimer Please consider Gantt repository as a no-op repo, without support. At the moment, this is more likely a sandbox than an active project. As said previously, any changes should happen in Nova, not in Gantt. As a gantt-core, I'm trying to follow all changes related to the scheduler in Nova (thanks to any commit msg having 'scheduler' in it) so as to make sure both Nova and forklift changes are going into the same direction. -Sylvain Susanne On Wed, Apr 9, 2014 at 12:44 PM, Henrique Truta henriquecostatr...@gmail.com wrote: @Oleg, @Sylvain, @Leandro, Thanls. I'll check the Gantt project and the blueprint 2014-04-09 12:59 GMT-03:00 Sylvain Bauza sylvain.ba...@gmail.com: 2014-04-09 17:47 GMT+02:00 Jay Lau jay.lau@gmail.com: @Oleg, Till now, I'm not sure the target of Gantt, is it for initial placement policy or run time policy or both, can you help clarify? I don't want to talk on behalf of Oleg, but Gantt is targeted to be the forklift of the current Nova scheduler. So, a placement decision based on dynamic metrics would be worth it. That said, as Gantt is not targeted to be delivered until Juno at least (with Nova sched deprecated), I think any progress on a BP should target Nova with respect to the forklift efforts, so it would automatically be ported to Gantt once the actual fork would happen. -Sylvain Jay ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev -- -- Ítalo Henrique Costa Truta ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
