It?s the same cloud infrastructure but not the same process flow. Let?s assume for a moment you have an ONAP component which runs on 3 VMs for fault tolerance (that?s not what we have in the demo setup but what you would have in production). In an ideal world you should be able to kill one of the VMs, spin up another one of the same type and the system should be happy again.
Today in reality you will have to go into the VM (sometimes more then one) after you spin it up and run various commands to ensure that new VM will be properly integrated with the 2 other once as well as ONAP overall. Not only do you have to run some commands but you will have to run different once depending on the component. Same is true for auto scaling. So as from the cloud layer it?s all the same the cloud layer is not enough. The way to fix that of course is to improve the code in each component so that this becomes fully automatic. The lack of that is a known shortcoming of the current implementation. Oliver On Apr 20, 2017, at 8:08 AM, Dhananjay Pavgi <DP00476350 at TechMahindra.com<mailto:DP00476350 at TechMahindra.com>> wrote: Thanks, Oliver. Absolutely got 1. Below and agreed. Didn?t get 2. Though, i.e. how every component is different when it comes to resiliency features. Ultimately, if it?s same underlying cloud infrastructure, right? thanks & regards, Dhananjay Pavgi Mobile : +91 98220 22264 <image001.png> <image002.jpg> www.techmahindra.com<https://urldefense.proofpoint.com/v2/url?u=http-3A__www.techmahindra.com_&d=DwMGaQ&c=LFYZ-o9_HUMeMTSQicvjIg&r=9iyuArzgyekj47PZSPfIijI2cSHsUJtAlcTA0X_udNI&m=N4OP98WhAr1is_7C1Yuy85OhdfcRxPfpekSCn4dhLFE&s=8JDxdisO-ASwqCdy1hE8QP4W_K7KhIdpyFMvSdylENk&e=> Platinum Member. Visit : http://www.onap.org<https://urldefense.proofpoint.com/v2/url?u=http-3A__www.onap.org_&d=DwMGaQ&c=LFYZ-o9_HUMeMTSQicvjIg&r=9iyuArzgyekj47PZSPfIijI2cSHsUJtAlcTA0X_udNI&m=N4OP98WhAr1is_7C1Yuy85OhdfcRxPfpekSCn4dhLFE&s=WgWBD8QmloguLRSmm6kAScTp_4qyPU1BwtdUIzZl4T0&e=> From: SPATSCHECK, OLIVER (OLIVER) [mailto:spat...@research.att.com] Sent: Thursday, April 20, 2017 5:04 PM To: Dhananjay Pavgi <DP00476350 at TechMahindra.com<mailto:DP00476350 at TechMahindra.com>> Cc: LEFEVRE, CATHERINE <cl664y at intl.att.com<mailto:cl664y at intl.att.com>>; onap-discuss at lists.onap.org<mailto:onap-discuss at lists.onap.org> Subject: Re: [onap-discuss] Support ONAP on multiple OS/Openstacks? Dhananjay, I think there are two dimensions ONAP can be improved in this regard. 1. Multi cloud support: To support multiple cloud technologies we not only have to modify the startup automation (heat template) but also adjust the components which work directly with the cloud orchestrator which are MSO, DCAE, App-C and Robotframework for every cloud technology we onboard. So to manage that effectively we should introduce a mediation layer between those 4 components and the underlying cloud technology. There are multiple options out there so that?s not something we have to build from scratch just something we have to pick and integrate. 2. The individual components are still very ?special?. What I mean with that is that each component has it?s own way to provide resilience, disaster recovery and scaling. So if you want to scale out or fail over e.g. the SDN-C controller you do something different then when you scale out DCAE etc? Those difference are visible on the cloud layer which makes this more difficult then it has to be. So we should figure out how to make this more homogeneous. Oliver On Apr 20, 2017, at 4:53 AM, Dhananjay Pavgi <DP00476350 at TechMahindra.com<mailto:DP00476350 at TechMahindra.com>> wrote: Fully agree, Catherine. In addition, should we also consider making ONAP agnostic of underlying virtualization layer e.g. OpenStack, VmWare, OpenVIM etc . Ideally, if cloud native approach is used then this should be taken care of. thanks & regards, Dhananjay Pavgi Mobile : +91 98220 22264 <image001.png> <image002.jpg> www.techmahindra.com<https://urldefense.proofpoint.com/v2/url?u=http-3A__www.techmahindra.com_&d=DwMFAw&c=LFYZ-o9_HUMeMTSQicvjIg&r=9iyuArzgyekj47PZSPfIijI2cSHsUJtAlcTA0X_udNI&m=qY5ElmRuugbIVM9UQVmuBI0vRH_tUKCg0HoRrqxrF2I&s=1yNUJKatHXXDvoRmPtigT-ey1sVtGVIVilkKUH7w9JU&e=> Platinum Member. Visit : http://www.onap.org<https://urldefense.proofpoint.com/v2/url?u=http-3A__www.onap.org_&d=DwMFAw&c=LFYZ-o9_HUMeMTSQicvjIg&r=9iyuArzgyekj47PZSPfIijI2cSHsUJtAlcTA0X_udNI&m=qY5ElmRuugbIVM9UQVmuBI0vRH_tUKCg0HoRrqxrF2I&s=nxL8tQrJ68d4Vg2TdSyRE-2cVoOgnc3SyZhebtGK40w&e=> From: onap-discuss-bounces at lists.onap.org<mailto:onap-discuss-bounces at lists.onap.org> [mailto:onap-discuss-boun...@lists.onap.org] On Behalf Of Lefevre, Catherine Sent: Thursday, April 20, 2017 2:11 PM To: onap-discuss at lists.onap.org<mailto:onap-discuss at lists.onap.org> Subject: [onap-discuss] Support ONAP on multiple OS/Openstacks? Good morning all, I would like to start a new thread in order to understand if there is a need to certify ONAP on multiple OS/Openstacks. OS Currently ONAP is running on Ubuntu 14.04 (target to move to Ubuntu 16.04). Open-O is running on CentOS but I understand from Helen Yunxia (Huawei) that it should not be an issue to run it on Ubuntu. Redhat was not considered from a certification perspective due to license cost.