Hi Alec, Yes, a reference to the paper you mentioned below would be nice:
[alec] we did not have to pay for the use of packet.net. as that work was done in the context of CNCF with participation from packet.net last year and lasted a few months (or perhaps CNCF negotiated a special rate). That was actually the first time that an open source traffic generator tool (let alone OPNFV tool) was automated to benchmark k8s based CNFs. CNCF published the results in a document (don’t have it OTOH but I can find a pointer if needed). OPNFV did not do any announcement on that as I was just too busy to do this kind of PR stuff. Even the title or enough info to narrow down a search would be helpful. thanks. Also: [alec] Lincoln has started an email thread to which I have provided more detailed information. Anybody is welcome to join the thread if we had proper mailing alias, or we could use a LaaS slack channel if there is one. there should be a “[LaaS]” Subject tag (or something like that) for the opnfv-tech-discuss list. IDK whether LaaS has a slack channel or not. It can be a problem to enable too many communication channels; projects loose info and focus – something we learned in OPNFV years ago. Al From: [email protected] [mailto:[email protected]] On Behalf Of Alec Hothan via lists.opnfv.org Sent: Friday, August 14, 2020 2:41 PM To: Jack Morgan <[email protected]>; opnfv-tech-discuss <[email protected]> Subject: Re: [opnfv-tech-discuss] Hardware Resources in OPNFV Hi Jack, See inline… From: <[email protected]<mailto:[email protected]>> on behalf of Jack Morgan <[email protected]<mailto:[email protected]>> Date: Friday, August 14, 2020 at 10:27 AM To: opnfv-tech-discuss <[email protected]<mailto:[email protected]>> Subject: Re: [opnfv-tech-discuss] Hardware Resources in OPNFV Alec, On 8/13/20 8:23 AM, Alec Hothan (ahothan) wrote: [alec] Packet.net is pretty good as we have actually been able to validate my project on packet.net with some collaboration with their infra team. This is interesting. Was this something you negociated on your own or through Linux Foundation? It brings up two points: 1) OPNFV projects looking to obtain their own hardware resources instead of working within the community (OPNFV Infra WG for example) 2) If you are using packet.net on your own instead of through Linux Foundation, I think we could potentially loose the ability to gain economies of scale in pricing or features requests as I'm sure there are other projects within Linux Foundation that could leverage using packet.net. [alec] we did not have to pay for the use of packet.net. as that work was done in the context of CNCF with participation from packet.net last year and lasted a few months (or perhaps CNCF negotiated a special rate). That was actually the first time that an open source traffic generator tool (let alone OPNFV tool) was automated to benchmark k8s based CNFs. CNCF published the results in a document (don’t have it OTOH but I can find a pointer if needed). OPNFV did not do any announcement on that as I was just too busy to do this kind of PR stuff. That project was pretty successful as we were almost within reach of providing a self-service SaaS high-performance traffic generator on packet.net. Involvement with the packet.net infar team was pretty minimal as our requirements were very basic (mainly has to deal with managing the pool of VLANs and configuring the relevant switch ports). Clearly nothing LaaS cannot do. IMHO any NFVi automated lab will require such basic logistical APIs (be it for tools or openstack or k8s). I was actually a bit surprised that the PDF discussion did not include such basic infra services. My hope is that I can use LaaS as a way to automate the CI for my project as a first step. Until now, validation has been done off OPNFV lab on resources of participating contributors. To Cedric’s point, I don’t really care if the LaaS are reserved statically or on demand, as long as I can run my CI on it with reasonable waiting time, I’m happy. Besides the being able to run CI well, I think the point we want to raise is if this is acceptable for supporting releases. I don't think LaaS has been used in this way in the past - as far as I know. [alec] well what is the exact definition of “OPNFV release support”, if it is to support any OPNFV project that participates in an OPNFV release, then the capabilities of LaaS are specific to each project. NFVbench requires a minimum LaaS setting to do CI gating of one part of the NFVbench code (that part that is basically independent of the NFVi cloud under test), it requires a much more advanced LaaS setting in order to test all advanced NFVbench features (mostly a working openstack deployment with several variations of network configuration/encapsulation). These advanced test can only be done in contributor ‘s labs today. To do that with LaaS, I would therefore need a full stack to test (be it openstack based or k8s based). Once LaaS can host any RI, it should not be too hard to automate the data plane benchmarking of that RI using NFVbench like it is being done in a lot of labs outside of OPNFV - this can be integrated in Functest, we’re just missing the LaaS resource requirements/reservation part. I think this is good feedback to the LaaS team. Thanks for you input. [alec] Lincoln has started an email thread to which I have provided more detailed information. Anybody is welcome to join the thread if we had proper mailing alias, or we could use a LaaS slack channel if there is one. Thanks Alec
-=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#24322): https://lists.opnfv.org/g/opnfv-tech-discuss/message/24322 Mute This Topic: https://lists.opnfv.org/mt/76157120/21656 Group Owner: [email protected] Unsubscribe: https://lists.opnfv.org/g/opnfv-tech-discuss/unsub [[email protected]] -=-=-=-=-=-=-=-=-=-=-=-
