[Engine-devel] Proposed next-generation vdsm API
Recently we've had some very productive discussions concerning the VDSM API. I want to attempt to refocus the discussion around an emerging proposal and see if we can agree on a sensible path forward. Based on the discussion, I have identified the following requirements that a new API for vdsm should have: 1.) Single API that can be consumed by ovirt-engine and ISVs - We don't want to maintain multiple parallel APIs - To develop a vendor ecosystem, we must have a robust external API to vdsm 2.) Full vdsm capabilities are exposed without requiring ovirt-engine - ovirt components should be modular and independently useful - Some deployments might want to manage nodes without ovirt-engine 3.) Standardized protocol with low overhead - Required for widespread adoption 4.) Support for asynchronous tasks and events - Needed by ovirt-engine and other consumers Based on these requirements, the following proposal has started to emerge: Create a REST API that will provide all of the functionality that is currently available via the xmlrpc interface (with the goal of deprecating xmlrpc once it becomes mature enough). To support advanced clustering features that ovirt-engine is planning, we'll write an QMF broker that can proxy the REST API onto a message bus. ovirt-engine will interact with vdsm exclusively over this bus but the REST API will be the principle API and the entry point for ISV apps. A REST API provides a light-weight and standard way to access all of the vdsm functionality. The REST API will handle events by exposing a new 'events' collection at the api root. REST users will use some sort of polling to collect these events. The details of this interface are being worked on. Several ways for minimizing the impact of polling have been discussed. The QMF broker can expose a publish/subscribe model for events as appropriate. Is this model an acceptable way to improve the vdsm API? I would like to hear the opinions of ovirt-engine developers, vdsm developers, and other stakeholders. Thanks for providing feedback on this proposal! -- Adam Litke <agl@us.ibm.com> IBM Linux Technology Center
Hi, i think this makes sense, but i'm not a VDSM expert. I did want to point out one other point, below: On 11/30/2011 11:40 PM, Adam Litke wrote:
Recently we've had some very productive discussions concerning the VDSM API. I want to attempt to refocus the discussion around an emerging proposal and see if we can agree on a sensible path forward.
Based on the discussion, I have identified the following requirements that a new API for vdsm should have:
1.) Single API that can be consumed by ovirt-engine and ISVs - We don't want to maintain multiple parallel APIs - To develop a vendor ecosystem, we must have a robust external API to vdsm
I have doubts around how useful the VDSM API will be for creating an ecosystem. If you look at most virtualization ISVs today, they want to integrate with a multi-node API and not a single-node API. The only use case that i know where integrating with a single node API is requested is when you're basically creating a virtualization management platform like oVirt itself. [Since we haven't met before, a brief intro... I have been responsible at Red Hat for buiding our virtualization ecosystem for the past year or so.] Regards, Geert
On Thu, Dec 01, 2011 at 12:37:00AM +0100, Geert Jansen wrote:
Hi,
i think this makes sense, but i'm not a VDSM expert. I did want to point out one other point, below:
On 11/30/2011 11:40 PM, Adam Litke wrote:
Recently we've had some very productive discussions concerning the VDSM API. I want to attempt to refocus the discussion around an emerging proposal and see if we can agree on a sensible path forward.
Based on the discussion, I have identified the following requirements that a new API for vdsm should have:
1.) Single API that can be consumed by ovirt-engine and ISVs - We don't want to maintain multiple parallel APIs - To develop a vendor ecosystem, we must have a robust external API to vdsm
I have doubts around how useful the VDSM API will be for creating an ecosystem. If you look at most virtualization ISVs today, they want to integrate with a multi-node API and not a single-node API. The only use case that i know where integrating with a single node API is requested is when you're basically creating a virtualization management platform like oVirt itself.
A single-node (or standalone VDSM deployment) is a very important use case. Many people are coming into the oVirt community from different perspectives. The strength of the ecosystem depends, in part, on the ability of oVirt components to be combined in unique ways with other software to produce solutions. The complete oVirt stack is a great thing, but not the only way to use the technology.
[Since we haven't met before, a brief intro... I have been responsible at Red Hat for buiding our virtualization ecosystem for the past year or so.]
Hi and thanks for the introduction! I look forward to working with you and the rest of the oVirt community on these issues :) -- Adam Litke <agl@us.ibm.com> IBM Linux Technology Center
On 12/01/2011 07:35 PM, Adam Litke wrote:
A single-node (or standalone VDSM deployment) is a very important use case. Many people are coming into the oVirt community from different perspectives. The strength of the ecosystem depends, in part, on the ability of oVirt components to be combined in unique ways with other software to produce solutions. The complete oVirt stack is a great thing, but not the only way to use the technology.
Just out of curiosity: what might those single node cases be, outside implementing an oVirt like solution? I think understanding those better is also critical input to the API design. Regards, Geert
On 12/01/2011 01:42 PM, Geert Jansen wrote:
On 12/01/2011 07:35 PM, Adam Litke wrote:
A single-node (or standalone VDSM deployment) is a very important use case. Many people are coming into the oVirt community from different perspectives. The strength of the ecosystem depends, in part, on the ability of oVirt components to be combined in unique ways with other software to produce solutions. The complete oVirt stack is a great thing, but not the only way to use the technology.
Just out of curiosity: what might those single node cases be, outside implementing an oVirt like solution? I think understanding those better is also critical input to the API design.
I would personally really like to see a server distro that installs virtual appliances instead of packages. With that I could have a server with a DHCP\httpd\mail server\webdav\IPA appliances running on the host completely separate using qemu or even lxc. Having a BL and UI around that would be very convenient for small setups and home servers. VDSM would be handy for storage, network and policy configuration as well as tracking the appliances stats.
Geert _______________________________________________ Engine-devel mailing list Engine-devel@ovirt.org http://lists.ovirt.org/mailman/listinfo/engine-devel
On 12/01/2011 12:42 PM, Geert Jansen wrote:
On 12/01/2011 07:35 PM, Adam Litke wrote:
A single-node (or standalone VDSM deployment) is a very important use case. Many people are coming into the oVirt community from different perspectives. The strength of the ecosystem depends, in part, on the ability of oVirt components to be combined in unique ways with other software to produce solutions. The complete oVirt stack is a great thing, but not the only way to use the technology.
Just out of curiosity: what might those single node cases be, outside implementing an oVirt like solution?
ovirt-engine currently doesn't handle every possible scenario where you want to manage more than one physical machine. And while I'm sure total world domination is not that far away, there's going to be a time period where there continues to be use cases it is not suited for. To name a few: 1) Fixed environments where a handful of systems are needed with some level of scripting used to coordinate things. Using a full three tier management server would be too much for this environment. This isn't necessarily a small deployment, this may be a huge number of small deployments (think half a dozen blades in the back of a retail store times 5,000). 2) Massive scale clusters. I'm talking Top 500 scale. There are people out there doing this with KVM today. They use tools like xCat and write directly against libvirt. But libvirt's lack of policy makes this harder than it should be. 3) Environments where virtualization is not the primary workload. There are a lot of cloud-like environments that are built with physical hardware only. In these cases, there is an existing infrastructure that does a lot of what oVirt does. It's easier for these environments to treat VMs as physical machines. As much as it's important to focus on the top-down view of oVirt as a cohesive stack, it's also important to look at each layer and make sure that each layer stands on its own. It's an 90/10 thing. You need a strong node-level interface to cover that last 10% of use-cases unless you're willing to spend 90% of the effort trying to also accommodate them. Regards, Anthony Liguori
I think understanding those better is also critical input to the API design.
Regards, Geert _______________________________________________ vdsm-devel mailing list vdsm-devel@lists.fedorahosted.org https://fedorahosted.org/mailman/listinfo/vdsm-devel
* Geert Jansen <gjansen@redhat.com> [2011-11-30 17:38]:
Hi,
i think this makes sense, but i'm not a VDSM expert. I did want to point out one other point, below:
On 11/30/2011 11:40 PM, Adam Litke wrote:
Recently we've had some very productive discussions concerning the VDSM API. I want to attempt to refocus the discussion around an emerging proposal and see if we can agree on a sensible path forward.
Based on the discussion, I have identified the following requirements that a new API for vdsm should have:
1.) Single API that can be consumed by ovirt-engine and ISVs - We don't want to maintain multiple parallel APIs - To develop a vendor ecosystem, we must have a robust external API to vdsm
I have doubts around how useful the VDSM API will be for creating an ecosystem. If you look at most virtualization ISVs today, they want to integrate with a multi-node API and not a single-node API. The only use case that i know where integrating with a single node API is requested is when you're basically creating a virtualization management platform like oVirt itself.
Without a first-class node level API, we're precluding the very case you're aware of. If we're building a community and ecosystem around KVM management then we need to be open to someone building that management platform and doing it in a way that keeps things compatible. There are existing products (IBM has a number in this space) which utilize libvirt as a node-level API which won't be able to (easily) integrate all of oVirt just to obtain access to the nodes. Further, if the only way to consume VDSM is via the multi-node solution, then price of entry is a much larger, more complex stack with more dependencies. This raises the barrier of entry and participation. IMHO, not ideal when we're attempting to grow a community. Alternatively, if we enable a common node API, not only do we support single node deployments (think virtual appliances, or hardware appliances; IBM has a few of these) but allowing competing (but compatible) multi-node/cluster solutions; and it even supports the single node solutions to be managed by different kvm-based management platforms because they're all working with the same node-level API. Having a first-class node API is a critical starting point for building our larger kvm management community since it allows for easier integration of existing products. I cannot stress that point enough; if we're committed to being an open community then we need design our interfaces to encourage participation. Lowing the cost of participation and integrate is great way to enable an ecosystem. -- Ryan Harper Software Engineer; Linux Technology Center IBM Corp., Austin, Tx ryanh@us.ibm.com
----- Original Message -----
* Geert Jansen <gjansen@redhat.com> [2011-11-30 17:38]:
Hi,
i think this makes sense, but i'm not a VDSM expert. I did want to point out one other point, below:
On 11/30/2011 11:40 PM, Adam Litke wrote:
Recently we've had some very productive discussions concerning the VDSM API. I want to attempt to refocus the discussion around an emerging proposal and see if we can agree on a sensible path forward.
Based on the discussion, I have identified the following requirements that a new API for vdsm should have:
1.) Single API that can be consumed by ovirt-engine and ISVs - We don't want to maintain multiple parallel APIs - To develop a vendor ecosystem, we must have a robust external API to vdsm
I have doubts around how useful the VDSM API will be for creating an ecosystem. If you look at most virtualization ISVs today, they want to integrate with a multi-node API and not a single-node API. The only use case that i know where integrating with a single node API is requested is when you're basically creating a virtualization management platform like oVirt itself.
Without a first-class node level API, we're precluding the very case you're aware of. If we're building a community and ecosystem around KVM management then we need to be open to someone building that management platform and doing it in a way that keeps things compatible.
There are existing products (IBM has a number in this space) which utilize libvirt as a node-level API which won't be able to (easily) integrate all of oVirt just to obtain access to the nodes.
Further, if the only way to consume VDSM is via the multi-node solution, then price of entry is a much larger, more complex stack with more dependencies. This raises the barrier of entry and participation. IMHO, not ideal when we're attempting to grow a community.
Alternatively, if we enable a common node API, not only do we support single node deployments (think virtual appliances, or hardware appliances; IBM has a few of these) but allowing competing (but compatible) multi-node/cluster solutions; and it even supports the single node solutions to be managed by different kvm-based management platforms because they're all working with the same node-level API.
Having a first-class node API is a critical starting point for building our larger kvm management community since it allows for easier integration of existing products. I cannot stress that point enough; if we're committed to being an open community then we need design our interfaces to encourage participation. Lowing the cost of participation and integrate is great way to enable an ecosystem.
+1 The REST-API should be exactly this. For multi-node environments where things are already complex we can raise the bar of demands, but for single node it has to be simple and straight-forward with very few requirements (i.e. requiring amqp in single-node use case is too complicated).
-- Ryan Harper Software Engineer; Linux Technology Center IBM Corp., Austin, Tx ryanh@us.ibm.com
_______________________________________________ vdsm-devel mailing list vdsm-devel@lists.fedorahosted.org https://fedorahosted.org/mailman/listinfo/vdsm-devel
-----Original Message----- From: Adam Litke [mailto:agl@us.ibm.com] Sent: Thursday, December 01, 2011 0:41 AM To: vdsm-devel@lists.fedorahosted.org; engine-devel@ovirt.org Cc: Daniel P. Berrange; Chris Wright; Dan Kenigsberg; Itamar Heim Subject: Proposed next-generation vdsm API
Recently we've had some very productive discussions concerning the VDSM API. I want to attempt to refocus the discussion around an emerging proposal and see if we can agree on a sensible path forward.
Based on the discussion, I have identified the following requirements
a new API for vdsm should have:
1.) Single API that can be consumed by ovirt-engine and ISVs - We don't want to maintain multiple parallel APIs - To develop a vendor ecosystem, we must have a robust external API to vdsm
2.) Full vdsm capabilities are exposed without requiring ovirt-engine - ovirt components should be modular and independently useful - Some deployments might want to manage nodes without ovirt-engine
3.) Standardized protocol with low overhead - Required for widespread adoption
4.) Support for asynchronous tasks and events - Needed by ovirt-engine and other consumers
Based on these requirements, the following proposal has started to emerge:
Create a REST API that will provide all of the functionality that is currently available via the xmlrpc interface (with the goal of deprecating xmlrpc once it becomes mature enough). To support advanced clustering features that ovirt-engine is planning, we'll write an QMF broker that can proxy the REST API onto a message bus. ovirt-engine will interact with vdsm exclusively over this bus but the REST API will be the principle API and the entry point for ISV apps. A REST API provides a light-weight and standard way to access all of the vdsm functionality.
The REST API will handle events by exposing a new 'events' collection at
that the api
root. REST users will use some sort of polling to collect these events. The details of this interface are being worked on. Several ways for minimizing the impact of polling have been discussed. The QMF broker can expose a publish/subscribe model for events as appropriate.
Is this model an acceptable way to improve the vdsm API? I would like to hear the opinions of ovirt-engine developers, vdsm developers, and other stakeholders. Thanks for providing feedback on this proposal!
Why things non native to REST and wrap it in QMF, rather than do the reverse? Or just to them in parallel, since it sounds like both are going to be first class citizens?
----- Original Message -----
-----Original Message----- From: Adam Litke [mailto:agl@us.ibm.com] Sent: Thursday, December 01, 2011 0:41 AM To: vdsm-devel@lists.fedorahosted.org; engine-devel@ovirt.org Cc: Daniel P. Berrange; Chris Wright; Dan Kenigsberg; Itamar Heim Subject: Proposed next-generation vdsm API
Recently we've had some very productive discussions concerning the VDSM API. I want to attempt to refocus the discussion around an emerging proposal and see if we can agree on a sensible path forward.
Based on the discussion, I have identified the following requirements that a new API for vdsm should have:
1.) Single API that can be consumed by ovirt-engine and ISVs - We don't want to maintain multiple parallel APIs - To develop a vendor ecosystem, we must have a robust external API to vdsm
2.) Full vdsm capabilities are exposed without requiring ovirt-engine - ovirt components should be modular and independently useful - Some deployments might want to manage nodes without ovirt-engine
3.) Standardized protocol with low overhead - Required for widespread adoption
4.) Support for asynchronous tasks and events - Needed by ovirt-engine and other consumers
Based on these requirements, the following proposal has started to emerge:
Create a REST API that will provide all of the functionality that is currently available via the xmlrpc interface (with the goal of deprecating xmlrpc once it becomes mature enough). To support advanced clustering features that ovirt-engine is planning, we'll write an QMF broker that can proxy the REST API onto a message bus. ovirt-engine will interact with vdsm exclusively over this bus but the REST API will be the principle API and the entry point for ISV apps. A REST API provides a light-weight and standard way to access all of the vdsm functionality.
The REST API will handle events by exposing a new 'events' collection at the api root. REST users will use some sort of polling to collect these events. The details of this interface are being worked on. Several ways for minimizing the impact of polling have been discussed. The QMF broker can expose a publish/subscribe model for events as appropriate.
Is this model an acceptable way to improve the vdsm API? I would like to hear the opinions of ovirt-engine developers, vdsm developers, and other stakeholders. Thanks for providing feedback on this proposal!
Why things non native to REST and wrap it in QMF, rather than do the reverse? Or just to them in parallel, since it sounds like both are going to be first class citizens?
This was more my understanding from our discussion on IRC yesterday. REST API - everything that is relevant for single node management QMF - same API as above + multi-node relevant API calls. I don't see any reason for doing weird things over REST to support the latter. In fact, I don't even see any real reason for going through the REST API when using QMF. If you take a look at today's API you will see that there is nothing there that limits it to XML-RPC and we could easily expose all the calls using REST or anything else. In python, exposing a new verb in the various APIs can be automatic so this would require very little maintenance. Any multi-node or transport specific calls can be decorated as such and would be automatically ignored/picked up by the relevant API layer. This way, we could also easily enable using different bus protocols assuming a customer already has a deployment as was suggested yesterday.
_______________________________________________ vdsm-devel mailing list vdsm-devel@lists.fedorahosted.org https://fedorahosted.org/mailman/listinfo/vdsm-devel
On Thu, Dec 01, 2011 at 02:02:27AM -0500, Ayal Baron wrote:
----- Original Message -----
-----Original Message----- From: Adam Litke [mailto:agl@us.ibm.com] Sent: Thursday, December 01, 2011 0:41 AM To: vdsm-devel@lists.fedorahosted.org; engine-devel@ovirt.org Cc: Daniel P. Berrange; Chris Wright; Dan Kenigsberg; Itamar Heim Subject: Proposed next-generation vdsm API
Recently we've had some very productive discussions concerning the VDSM API. I want to attempt to refocus the discussion around an emerging proposal and see if we can agree on a sensible path forward.
Based on the discussion, I have identified the following requirements that a new API for vdsm should have:
1.) Single API that can be consumed by ovirt-engine and ISVs - We don't want to maintain multiple parallel APIs - To develop a vendor ecosystem, we must have a robust external API to vdsm
2.) Full vdsm capabilities are exposed without requiring ovirt-engine - ovirt components should be modular and independently useful - Some deployments might want to manage nodes without ovirt-engine
3.) Standardized protocol with low overhead - Required for widespread adoption
4.) Support for asynchronous tasks and events - Needed by ovirt-engine and other consumers
Based on these requirements, the following proposal has started to emerge:
Create a REST API that will provide all of the functionality that is currently available via the xmlrpc interface (with the goal of deprecating xmlrpc once it becomes mature enough). To support advanced clustering features that ovirt-engine is planning, we'll write an QMF broker that can proxy the REST API onto a message bus. ovirt-engine will interact with vdsm exclusively over this bus but the REST API will be the principle API and the entry point for ISV apps. A REST API provides a light-weight and standard way to access all of the vdsm functionality.
The REST API will handle events by exposing a new 'events' collection at the api root. REST users will use some sort of polling to collect these events. The details of this interface are being worked on. Several ways for minimizing the impact of polling have been discussed. The QMF broker can expose a publish/subscribe model for events as appropriate.
Is this model an acceptable way to improve the vdsm API? I would like to hear the opinions of ovirt-engine developers, vdsm developers, and other stakeholders. Thanks for providing feedback on this proposal!
Why things non native to REST and wrap it in QMF, rather than do the reverse? Or just to them in parallel, since it sounds like both are going to be first class citizens?
This was more my understanding from our discussion on IRC yesterday.
I'm afraid I did not follow that discussion to your conclusions...
REST API - everything that is relevant for single node management QMF - same API as above + multi-node relevant API calls. I don't see any reason for doing weird things over REST to support the latter.
A QMF broker runs on the vdsm host and talks to the REST API. It connects to a bus and exposes an API to ovirt-engine on this bus using a vdsm-base schema. ovirt-engine wants additional clustering functionality. This API should be implemented completely between the QMF broker and ovirt-engine using a separate vdsm-cluster schema.
In fact, I don't even see any real reason for going through the REST API when using QMF.
Because we want to avoid the proliferation of APIs. I would prefer a mostly vertical chain of API components to a vdsm with several independent APIs (which are sure to diverge or be neglected by individual developers/patches).
If you take a look at today's API you will see that there is nothing there that limits it to XML-RPC and we could easily expose all the calls using REST or anything else. In python, exposing a new verb in the various APIs can be automatic so this would require very little maintenance. Any multi-node or transport specific calls can be decorated as such and would be automatically ignored/picked up by the relevant API layer. This way, we could also easily enable using different bus protocols assuming a customer already has a deployment as was suggested yesterday.
I don't think this will be as automatic as you suggest in practice. It sounds like it will increase code complexity, obfuscation, and maintenance burden. -- Adam Litke <agl@us.ibm.com> IBM Linux Technology Center
----- Original Message -----
On Thu, Dec 01, 2011 at 02:02:27AM -0500, Ayal Baron wrote:
----- Original Message -----
-----Original Message----- From: Adam Litke [mailto:agl@us.ibm.com] Sent: Thursday, December 01, 2011 0:41 AM To: vdsm-devel@lists.fedorahosted.org; engine-devel@ovirt.org Cc: Daniel P. Berrange; Chris Wright; Dan Kenigsberg; Itamar Heim Subject: Proposed next-generation vdsm API
Recently we've had some very productive discussions concerning the VDSM API. I want to attempt to refocus the discussion around an emerging proposal and see if we can agree on a sensible path forward.
Based on the discussion, I have identified the following requirements that a new API for vdsm should have:
1.) Single API that can be consumed by ovirt-engine and ISVs - We don't want to maintain multiple parallel APIs - To develop a vendor ecosystem, we must have a robust external API to vdsm
2.) Full vdsm capabilities are exposed without requiring ovirt-engine - ovirt components should be modular and independently useful - Some deployments might want to manage nodes without ovirt-engine
3.) Standardized protocol with low overhead - Required for widespread adoption
4.) Support for asynchronous tasks and events - Needed by ovirt-engine and other consumers
Based on these requirements, the following proposal has started to emerge:
Create a REST API that will provide all of the functionality that is currently available via the xmlrpc interface (with the goal of deprecating xmlrpc once it becomes mature enough). To support advanced clustering features that ovirt-engine is planning, we'll write an QMF broker that can proxy the REST API onto a message bus. ovirt-engine will interact with vdsm exclusively over this bus but the REST API will be the principle API and the entry point for ISV apps. A REST API provides a light-weight and standard way to access all of the vdsm functionality.
The REST API will handle events by exposing a new 'events' collection at the api root. REST users will use some sort of polling to collect these events. The details of this interface are being worked on. Several ways for minimizing the impact of polling have been discussed. The QMF broker can expose a publish/subscribe model for events as appropriate.
Is this model an acceptable way to improve the vdsm API? I would like to hear the opinions of ovirt-engine developers, vdsm developers, and other stakeholders. Thanks for providing feedback on this proposal!
Why things non native to REST and wrap it in QMF, rather than do the reverse? Or just to them in parallel, since it sounds like both are going to be first class citizens?
This was more my understanding from our discussion on IRC yesterday.
I'm afraid I did not follow that discussion to your conclusions...
REST API - everything that is relevant for single node management QMF - same API as above + multi-node relevant API calls. I don't see any reason for doing weird things over REST to support the latter.
A QMF broker runs on the vdsm host and talks to the REST API. It connects to a bus and exposes an API to ovirt-engine on this bus using a vdsm-base schema. ovirt-engine wants additional clustering functionality. This API should be implemented completely between the QMF broker and ovirt-engine using a separate vdsm-cluster schema.
In fact, I don't even see any real reason for going through the REST API when using QMF.
Because we want to avoid the proliferation of APIs. I would prefer a mostly vertical chain of API components to a vdsm with several independent APIs (which are sure to diverge or be neglected by individual developers/patches).
If you take a look at today's API you will see that there is nothing there that limits it to XML-RPC and we could easily expose all the calls using REST or anything else. In python, exposing a new verb in the various APIs can be automatic so this would require very little maintenance. Any multi-node or transport specific calls can be decorated as such and would be automatically ignored/picked up by the relevant API layer. This way, we could also easily enable using different bus protocols assuming a customer already has a deployment as was suggested yesterday.
I don't think this will be as automatic as you suggest in practice. It sounds like it will increase code complexity, obfuscation, and maintenance burden.
Even today we don't really have any reliance on the transport. The beauty of REST is its simplicity, but with that come limitations. If we require the qmf API to go through REST then we're basically limiting the qmf to the REST shortcomings or we implement things that bend REST the wrong way to accomodate the extra requirements. For any "simple" call I'm fine with qmf going through REST if it simplifies maintenance, but for things like events, I believe there is no justification for going through REST.
-- Adam Litke <agl@us.ibm.com> IBM Linux Technology Center
On Wed, Nov 30, 2011 at 11:32:55PM -0500, Itamar Heim wrote:
-----Original Message----- From: Adam Litke [mailto:agl@us.ibm.com] Sent: Thursday, December 01, 2011 0:41 AM To: vdsm-devel@lists.fedorahosted.org; engine-devel@ovirt.org Cc: Daniel P. Berrange; Chris Wright; Dan Kenigsberg; Itamar Heim Subject: Proposed next-generation vdsm API
Recently we've had some very productive discussions concerning the VDSM API. I want to attempt to refocus the discussion around an emerging proposal and see if we can agree on a sensible path forward.
Based on the discussion, I have identified the following requirements that a new API for vdsm should have:
1.) Single API that can be consumed by ovirt-engine and ISVs - We don't want to maintain multiple parallel APIs - To develop a vendor ecosystem, we must have a robust external API to vdsm
2.) Full vdsm capabilities are exposed without requiring ovirt-engine - ovirt components should be modular and independently useful - Some deployments might want to manage nodes without ovirt-engine
3.) Standardized protocol with low overhead - Required for widespread adoption
4.) Support for asynchronous tasks and events - Needed by ovirt-engine and other consumers
Based on these requirements, the following proposal has started to emerge:
Create a REST API that will provide all of the functionality that is currently available via the xmlrpc interface (with the goal of deprecating xmlrpc once it becomes mature enough). To support advanced clustering features that ovirt-engine is planning, we'll write an QMF broker that can proxy the REST API onto a message bus. ovirt-engine will interact with vdsm exclusively over this bus but the REST API will be the principle API and the entry point for ISV apps. A REST API provides a light-weight and standard way to access all of the vdsm functionality.
The REST API will handle events by exposing a new 'events' collection at the api root. REST users will use some sort of polling to collect these events. The details of this interface are being worked on. Several ways for minimizing the impact of polling have been discussed. The QMF broker can expose a publish/subscribe model for events as appropriate.
Is this model an acceptable way to improve the vdsm API? I would like to hear the opinions of ovirt-engine developers, vdsm developers, and other stakeholders. Thanks for providing feedback on this proposal!
Why things non native to REST and wrap it in QMF, rather than do the reverse? Or just to them in parallel, since it sounds like both are going to be first class citizens?
Parallel APIs mean dual maintenance. There will be inherent incompatibilities as each API would naturally have small differences. The reason for beginning with REST because of its low overhead and simplicity. Users of the REST API would not need to concern themselves with QMF at all but if that extra set of features is desired it can be easily added. -- Adam Litke <agl@us.ibm.com> IBM Linux Technology Center
On 11/30/2011 05:40 PM, Adam Litke wrote:
Recently we've had some very productive discussions concerning the VDSM API. I want to attempt to refocus the discussion around an emerging proposal and see if we can agree on a sensible path forward.
Based on the discussion, I have identified the following requirements that a new API for vdsm should have:
1.) Single API that can be consumed by ovirt-engine and ISVs - We don't want to maintain multiple parallel APIs - To develop a vendor ecosystem, we must have a robust external API to vdsm
2.) Full vdsm capabilities are exposed without requiring ovirt-engine - ovirt components should be modular and independently useful - Some deployments might want to manage nodes without ovirt-engine
3.) Standardized protocol with low overhead - Required for widespread adoption
4.) Support for asynchronous tasks and events - Needed by ovirt-engine and other consumers
Based on these requirements, the following proposal has started to emerge:
Create a REST API that will provide all of the functionality that is currently available via the xmlrpc interface I just want to stress that it WILL deprecate and significantly change some aspects of the API most obvious are the task aspects of the API. Even though the commands might be the same the current async\sync task semantics might change to better accommodate the QMF wrapper and other users. I don't think it's needed to go into details now but I am stressing that there WILL be API calls the will not migrate or significantly change in the REST API. (with the goal of deprecating xmlrpc once it becomes mature enough). To support advanced clustering features that ovirt-engine is planning, we'll write an QMF broker that can proxy the REST API onto a message bus. ovirt-engine will interact with vdsm exclusively over this bus but the REST API will be the principle API and the entry point for ISV apps. A REST API provides a light-weight and standard way to access all of the vdsm functionality.
The REST API will handle events by exposing a new 'events' collection at the api root. REST users will use some sort of polling to collect these events. The details of this interface are being worked on. Several ways for minimizing the impact of polling have been discussed. The QMF broker can expose a publish/subscribe model for events as appropriate.
Is this model an acceptable way to improve the vdsm API? I would like to hear the opinions of ovirt-engine developers, vdsm developers, and other stakeholders. Thanks for providing feedback on this proposal!
participants (7)
-
Adam Litke -
Anthony Liguori -
Ayal Baron -
Geert Jansen -
Itamar Heim -
Ryan Harper -
Saggi Mizrahi