Re: Subject: Re: [Trove] State of the Trove service tenant deployment model
On Wed, Jan 23, 2019 at 9:27 AM Fox, Kevin M <Kevin.Fox@pnnl.gov<mailto:Kevin.Fox@pnnl.gov>> wrote:
I'd recommend at this point to maybe just run kubernetes across the vms and push the guest agents/workload to them.
This sounds like an overkill to me. Currently, different projects in openstack are solving this issue > in different ways, e.g. Octavia is using two-way SSL authentication API between the controller service and amphora(which is the vm running HTTP server inside), Magnum is using heat-container-agent that is communicating with Heat via API, etc. However, Trove chooses another option which has brought a lot of discussions over a long time.
In the current situation, I don't think it's doable for each project heading to one common solution, but Trove can learn from other projects to solve its own problem. Cheers, Lingxian Kong
The Octavia way of communication was discussed by Trove several times in the context of secuirty. However, the security threat has been eliminated by encryption. I'm wondering if the Octavia way prevents DDOS attacks also ? Implementation of two-way SSL authentication API could be included in the Trove priority list IMHO if it solves all issues with security/DDOS attacks. This could also creates some share code between both projects and help other services as well. Best, Darek
Sorry for bringing this thread back to the top again. But I am wondering if there are people who have already deployed Trove in production? If yes, are you using service tenant model(create the database vm and related resources in the admin project) or using the flatten mode that the end user has access to the database vm and the control plane network as well? I am asking because we are going to deploy Trove in a private cloud, and we want to take more granular control of the resources created, e.g for every database vm, we will create the vm in the admin tenant, plug a port to the control plane(`CONF.default_neutron_networks`) and the other ports to the network given by the users, we also need to specify different security groups to different types of neutron ports for security reasons, etc. There are something missing in trove in order to achieve the above, I'm working on that, but I'd like to hear more suggestions. My irc name is lxkong in #openstack-trove, please ping me if you have something to share. Cheers, Lingxian Kong On Wed, Jan 23, 2019 at 7:35 PM Darek Król <dkrol3@gmail.com> wrote:
On Wed, Jan 23, 2019 at 9:27 AM Fox, Kevin M <Kevin.Fox@pnnl.gov<mailto:Kevin.Fox@pnnl.gov>> wrote:
I'd recommend at this point to maybe just run kubernetes across the vms and push the guest agents/workload to them.
This sounds like an overkill to me. Currently, different projects in openstack are solving this issue > in different ways, e.g. Octavia is using two-way SSL authentication API between the controller service and amphora(which is the vm running HTTP server inside), Magnum is using heat-container-agent that is communicating with Heat via API, etc. However, Trove chooses another option which has brought a lot of discussions over a long time.
In the current situation, I don't think it's doable for each project heading to one common solution, but Trove can learn from other projects to solve its own problem. Cheers, Lingxian Kong
The Octavia way of communication was discussed by Trove several times in the context of secuirty. However, the security threat has been eliminated by encryption. I'm wondering if the Octavia way prevents DDOS attacks also ?
Implementation of two-way SSL authentication API could be included in the Trove priority list IMHO if it solves all issues with security/DDOS attacks. This could also creates some share code between both projects and help other services as well.
Best, Darek
Hello Lingxian, I’ve heard about a few tries of running Trove in production. Unfortunately, I didn’t have opportunity to get details about networking. At Samsung, we introducing Trove into our products for on-premise cloud platforms. However, I cannot share too many details about it, besides it is oriented towards performance and security is not a concern. Hence, the networking is very basic without any layers of abstractions if possible. Could you share more details about your topology and goals you want to achieve in Trove ? Maybe Trove team could help you in this ? Unfortunately, I’m not a network expert so I would need to get more details to understand your use case better. I would also like to get this opportunity to ask you for details about Octavia way of communication ? I'm wondering if the Octavia way prevents DDOS attacks also ? Best, Darek On Fri, 8 Feb 2019 at 01:20, Lingxian Kong <anlin.kong@gmail.com> wrote:
Sorry for bringing this thread back to the top again.
But I am wondering if there are people who have already deployed Trove in production? If yes, are you using service tenant model(create the database vm and related resources in the admin project) or using the flatten mode that the end user has access to the database vm and the control plane network as well?
I am asking because we are going to deploy Trove in a private cloud, and we want to take more granular control of the resources created, e.g for every database vm, we will create the vm in the admin tenant, plug a port to the control plane(`CONF.default_neutron_networks`) and the other ports to the network given by the users, we also need to specify different security groups to different types of neutron ports for security reasons, etc.
There are something missing in trove in order to achieve the above, I'm working on that, but I'd like to hear more suggestions.
My irc name is lxkong in #openstack-trove, please ping me if you have something to share.
Cheers, Lingxian Kong
On Wed, Jan 23, 2019 at 7:35 PM Darek Król <dkrol3@gmail.com> wrote:
On Wed, Jan 23, 2019 at 9:27 AM Fox, Kevin M <Kevin.Fox@pnnl.gov<mailto:Kevin.Fox@pnnl.gov>> wrote:
I'd recommend at this point to maybe just run kubernetes across the vms and push the guest agents/workload to them.
This sounds like an overkill to me. Currently, different projects in openstack are solving this issue > in different ways, e.g. Octavia is using two-way SSL authentication API between the controller service and amphora(which is the vm running HTTP server inside), Magnum is using heat-container-agent that is communicating with Heat via API, etc. However, Trove chooses another option which has brought a lot of discussions over a long time.
In the current situation, I don't think it's doable for each project heading to one common solution, but Trove can learn from other projects to solve its own problem. Cheers, Lingxian Kong
The Octavia way of communication was discussed by Trove several times in the context of secuirty. However, the security threat has been eliminated by encryption. I'm wondering if the Octavia way prevents DDOS attacks also ?
Implementation of two-way SSL authentication API could be included in the Trove priority list IMHO if it solves all issues with security/DDOS attacks. This could also creates some share code between both projects and help other services as well.
Best, Darek
On Sun, Feb 10, 2019 at 7:04 AM Darek Król <dkrol3@gmail.com> wrote:
Hello Lingxian,
I’ve heard about a few tries of running Trove in production. Unfortunately, I didn’t have opportunity to get details about networking. At Samsung, we introducing Trove into our products for on-premise cloud platforms. However, I cannot share too many details about it, besides it is oriented towards performance and security is not a concern. Hence, the networking is very basic without any layers of abstractions if possible.
Could you share more details about your topology and goals you want to achieve in Trove ? Maybe Trove team could help you in this ? Unfortunately, I’m not a network expert so I would need to get more details to understand your use case better.
Yeah, I think trove team could definitely help. I've been working on a patch[1] to support different sgs for different type of neutron ports, the patch is for the use case that `CONF.default_neutron_networks` is configured as trove management network. Besides, I also have some patches[2][3] for trove need to be reviewed, not sure who are the right people I should ask for review now, but would appriciate if you could help. [1]: https://review.openstack.org/#/c/635705/ [2]: https://review.openstack.org/#/c/635099/ [3]: https://review.openstack.org/#/c/635138/ Cheers, Lingxian Kong
Lingxian Kong wrote:
On Sun, Feb 10, 2019 at 7:04 AM Darek Król <dkrol3@gmail.com <mailto:dkrol3@gmail.com>> wrote:
Hello Lingxian,
I’ve heard about a few tries of running Trove in production. Unfortunately, I didn’t have opportunity to get details about networking. At Samsung, we introducing Trove into our products for on-premise cloud platforms. However, I cannot share too many details about it, besides it is oriented towards performance and security is not a concern. Hence, the networking is very basic without any layers of abstractions if possible.
Could you share more details about your topology and goals you want to achieve in Trove ? Maybe Trove team could help you in this ? Unfortunately, I’m not a network expert so I would need to get more details to understand your use case better.
Yeah, I think trove team could definitely help. I've been working on a patch[1] to support different sgs for different type of neutron ports, the patch is for the use case that `CONF.default_neutron_networks` is configured as trove management network.
Besides, I also have some patches[2][3] for trove need to be reviewed, not sure who are the right people I should ask for review now, but would appriciate if you could help.
I think OVH has been deploying Trove as well, or at least considering it... Ccing Jean-Daniel in case he can bring some insights on that. -- Thierry
Hi, We are mainly focus on Ironic, we put the DB topics a little bit aside. Sorry. Jean-Daniel Bonnetot ovh.com <http://ovh.com> | @pilgrimstack On 11/02/2019 16:51, "Thierry Carrez" <thierry@openstack.org> wrote: Lingxian Kong wrote: > On Sun, Feb 10, 2019 at 7:04 AM Darek Król <dkrol3@gmail.com > <mailto:dkrol3@gmail.com>> wrote: > > Hello Lingxian, > > I’ve heard about a few tries of running Trove in production. > Unfortunately, I didn’t have opportunity to get details about > networking. At Samsung, we introducing Trove into our products for > on-premise cloud platforms. However, I cannot share too many details > about it, besides it is oriented towards performance and security is > not a concern. Hence, the networking is very basic without any > layers of abstractions if possible. > > Could you share more details about your topology and goals you want > to achieve in Trove ? Maybe Trove team could help you in this ? > Unfortunately, I’m not a network expert so I would need to get more > details to understand your use case better. > > > Yeah, I think trove team could definitely help. I've been working on a > patch[1] to support different sgs for different type of neutron ports, > the patch is for the use case that `CONF.default_neutron_networks` is > configured as trove management network. > > Besides, I also have some patches[2][3] for trove need to be reviewed, > not sure who are the right people I should ask for review now, but would > appriciate if you could help. I think OVH has been deploying Trove as well, or at least considering it... Ccing Jean-Daniel in case he can bring some insights on that. -- Thierry
participants (4)
-
Darek Król
-
Jean-Daniel Bonnetot
-
Lingxian Kong
-
Thierry Carrez