Thanks a lot. A little more... During SDC design phase, we start out by creating a VSP ( Vendor Software Product ) and then import VSP to create VF ( Virtual Function ). Then using "Create Network Service ", we combine multiple VF to generate a Network Service ( NS ) in SDC, which are then certified, approved & distributed.
While NS is already created in design time, where there is an additional step of creating VNF ( Virtual Network Function ) in VID component. BR, Viswa On Wed, Jun 7, 2017 at 8:30 PM, ROSE, DANIEL V <[email protected]> wrote: > Hi, > > > > 1. It is a customer. I don’t know what NS is but aai does hold > relationships. > > 2. The model is very confusingly named amongst different systems. > Vf aka vfmodule is the smallest deployable piece of a vnf. Vnf is a single > virtual function. Service aka service instance can have multiple vnfs > > 3. Robot vm is not letting aai know about the asdc models, those > are done during distribution in the asdc gui. Aai listens to dmaap for info > on distributed models. Robot vm does have test cases to do them but you > don’t to do them through robot. Robot’s init command in demo.sh does need > to let aai know about open stack regions, tenants etc. That’s because those > are not created as part of an onap spin up. You could build a gui or > something to do it but its not something you need to do for every vnf > installation. > > 4. I attached the email where I described the cloud config file. > LCP Region is equivalent to openstack region ie something like IAD or DFW > in Rackspace. > > > > Thanks, > > Daniel Rose > > ECOMP / ONAP > > com.att.ecomp > > 732-420-7308 > > > > *From:* Viswa KSP [mailto:[email protected]] > *Sent:* Wednesday, June 07, 2017 9:35 AM > *To:* ROSE, DANIEL V <[email protected]> > *Cc:* Dhananjay Pavgi <[email protected]>; > [email protected] > > *Subject:* Re: [onap-discuss] Why there is a manual step before deploying > a distributed / certified network service? > > > > Thanks Daniel. Could you please clarify some of my follow-up queries: > > > > 1. Could you help me with the notion of "Subscribers" ? Why & for what > use-case? > > > - Would AAI hold this information of what subscribers are connected to > which instance of NS ? > - Can multiple subscribers share same NS? Would this information be > exported / connected to external systems like accounting / BSS ? > > > 1. Moreover, I would like to understand the difference between Virtual > Network Function ( during deploy time at VID ) and Virtual Function ( > during design time at SDC ). Why there is another layer called VNF on top > of VF ? > 2. Robot-VM is actually doing many stuffs including letting AAI know > about the SDC models and letting AAI know about Openstack tenant details ( > by using APIs ). Is there any WIP to automate the same? > > > - Why there is no communication b/w SDC and AAI, as soon a network > service / VF is certified & distributed by operator? > > > 1. On a separate thread, Catherine had mentioned that, there is a file > called cloud-config.json which can be used to specify regions / tenants > within openstack. I believe that is for MSO. If I want to deploy NS on > selected region / tenant in VID, how am I supposed to achieve it? > > > - I also see something termed as "LCP region". Could you please > explain more about this or Please point me to some resource in Wiki > where I > can read more about this. > > BR, > > Viswa > > > > > > > > On Thu, Jun 1, 2017 at 11:29 PM, ROSE, DANIEL V <[email protected]> wrote: > > To Dhananjay, > > > > 1. For configuration of devices yes, but the demos don’t use them > because the virtual functions are not configured they are hard coded. A > real vnf would need to have yang. > > 2. For the demos, nothing. > > 3. 3a is a one time thing, it does stuff like populate vim stuff > (region and tenant) the expectation in a real cloud is youd fill those in > as regions are created. It also creates a subscriber which is basically a > customer and would presumably be done upstream in a bss / oss / wherever > when they are first created. 3b, 3c and 3e I think there is an enhancement > in the works to allow the whole thing to be instantiated at once, mso can > confirm. It may already be possible, since we do have the model we know the > steps of everything. 3d is a bit of an odd child, its data that is needed > to complete the instantiation but which is not provided in the model. > Internally we use external systems for a lot of it (ipam etc) and the > preload is just what is left of that. The rebase we are testing now should > remove the need for this as this info is asked for in vid. 3f mso will do, > there is a story open for that. > > 4. 3b, 3c, 3e are already done in SO (nee MSO), 3a is outside > process (you don’t create tenants in openstack when calling nova create > server for instance) and 3d and 3f are temporary things that should be > going away soon. > > > > To Viswa, > > This decomposition is already done, just not shown in the demo so well, > the model and the SO do know the order. And MSO is involved (see response > to Dhananjay above). > > > > > > Thanks, > > Daniel Rose > > ECOMP / ONAP > > com.att.ecomp > > 732-420-7308 > > > > *From:* Viswa KSP [mailto:[email protected]] > *Sent:* Thursday, June 01, 2017 1:43 PM > *To:* Dhananjay Pavgi <[email protected]> > *Cc:* [email protected]; ROSE, DANIEL V <[email protected]> > *Subject:* RE: [onap-discuss] Why there is a manual step before deploying > a distributed / certified network service? > > > > Thanks Daniel. I do have a follow up query. Adding to what Dhananjay has > quoted, should the decomposition of network service be taken care > automatically by service orchestrator ? > > > > If we see Open -O or XOS/CORD, the story remains same . We provide NSD and > the master orchestrator takes the burden of decomposing network service > requirements, identify dependencies w.r.t infrastructure ( NF -O ) and > network ( SDN-O ) , along with life cycle work flow. > > > > Is there any reason ( architectururaly ) , why MSO isn't involved here ? > > > > BR, > > Viswa > > > > On Jun 1, 2017 9:47 PM, "Dhananjay Pavgi" <[email protected]> > wrote: > > Thanks, Daniel. Few related questions as below: > > > > 1. When it comes to “Service Instantiation” on VNFs; don’t we need > Yang models. > > 2. i.e. what does finally SDNC send as a payload over Netconf? > > 3. Also, our understanding of APIs being called is as below (which > is right now manual activity in vFW demo): > > a. Run demo.sh init script (We will share these details but > essentially it invokes AAI REST APIs) > > b. Service Instance Create > > c. VNF Create > > d. Run demo.sh preload (We will share these details but essentially > it invokes SDNC REST APIs) > > e. VF Module Create > > f. Run demo.sh appc script (We will share these details but > essentially it invokes APPC REST APIs) > > 4. Ideally, 3a through 3f above should be handled by “Service > Orchestrator”, right? > > > > thanks & regards, > > Dhananjay Pavgi > > Mobile : +91 98220 22264 > > [image: cid:[email protected]] [image: > ONAP_logo_Sig] > > www.techmahindra.com > <https://urldefense.proofpoint.com/v2/url?u=http-3A__www.techmahindra.com_&d=DwMFaQ&c=LFYZ-o9_HUMeMTSQicvjIg&r=2wwdGZ3YcpSivQ2Kio028A&m=wj-P9f9XBIxdpyIqrRsIg4InkRWIvIwz0MSTL3reH3M&s=eKSSa5svrJJZIBjHGiq0bOAK3KpUelMy_gIEFUd7aXM&e=> > Platinum Member. Visit : http://www.onap.org > <https://urldefense.proofpoint.com/v2/url?u=http-3A__www.onap.org_&d=DwMFaQ&c=LFYZ-o9_HUMeMTSQicvjIg&r=2wwdGZ3YcpSivQ2Kio028A&m=wj-P9f9XBIxdpyIqrRsIg4InkRWIvIwz0MSTL3reH3M&s=XGAJqN0REk91XmBABHIg5_3B4vOdE6XO9tZMP59_dZo&e=> > > > > *From:* [email protected] [mailto:onap-discuss-bounces@ > lists.onap.org] *On Behalf Of *ROSE, DANIEL V > *Sent:* Thursday, June 01, 2017 7:54 PM > *To:* Viswa KSP <[email protected]>; [email protected] > *Subject:* Re: [onap-discuss] Why there is a manual step before deploying > a distributed / certified network service? > > > > The demo is specifically just a demo. So the stuff in the demo like the > robot scripts will not work for other vnfs. The mechanisms used in the > scripts will work with any vnfs however. > > > > To give you an example, preload command calls sdnc rest api that loads > info like IPs to sdnc so that when you go to vid and instantiate a vnf sdnc > knows what data to use. You can call that api yourself to fill in your > specific parameters for ANY vnf ( you can add a call to the DG to make a > call out to any external system via rest instead of doing the preload > also). In the rebase code VID has the option to provide those preloads as > part of the GUI to make it a bit easier but even though the flow changes > the fact that it applies to all VNFs does not change. > > > > Thanks, > > Daniel Rose > > ECOMP / ONAP > > com.att.ecomp > > 732-420-7308 > > > > *From:* [email protected] [mailto:onap-discuss-bounces@ > lists.onap.org <[email protected]>] *On Behalf Of *Viswa > KSP > *Sent:* Thursday, June 01, 2017 4:50 AM > *To:* [email protected] > *Subject:* [onap-discuss] Why there is a manual step before deploying a > distributed / certified network service? > > > > Dear All, > > > > We are trying to follow the steps as per tutorial to bring demo vFirewall > Network service. > > While the steps before deploy operation ( i.e creation of license model, > design-test-certify cycle for VSP, VF & NS and finally distributing NS for > production ) seems to be generic & acceptable for other VNFs, the step for > running demo.sh in robot VM to enable "Deploy" in VID seems to very > specific to vFirewall NS. > > > > I would like to understand if this manual step is expected in any other > NS? What if we wanna customize the work-flow i.e say I want to just design > my NS and just deploy it right away ( without other steps as per demo > work-flow ) , how should I go-about in achieving this? > > > > On a very high-level, I would like to know how much we can customize ONAP ? > > > > BR, > > Viswa > > ============================================================ > ================================================================ > > Disclaimer: This message and the information contained herein is > proprietary and confidential and subject to the Tech Mahindra policy > statement, you may review the policy at http://www.techmahindra.com/ > Disclaimer.html > <https://urldefense.proofpoint.com/v2/url?u=http-3A__www.techmahindra.com_Disclaimer.html&d=DwMFaQ&c=LFYZ-o9_HUMeMTSQicvjIg&r=2wwdGZ3YcpSivQ2Kio028A&m=wj-P9f9XBIxdpyIqrRsIg4InkRWIvIwz0MSTL3reH3M&s=OA8rnrS-dBDPZLs0KDL3htkm1D9AsYCaCxZoSDkaViw&e=> > externally http://tim.techmahindra.com/tim/disclaimer.html > <https://urldefense.proofpoint.com/v2/url?u=http-3A__tim.techmahindra.com_tim_disclaimer.html&d=DwMFaQ&c=LFYZ-o9_HUMeMTSQicvjIg&r=2wwdGZ3YcpSivQ2Kio028A&m=wj-P9f9XBIxdpyIqrRsIg4InkRWIvIwz0MSTL3reH3M&s=mOiRMi1aNG1XHwE9Vf78B94NGLAGHOKf_Iy5mdLlchU&e=> > internally within TechMahindra. > > ============================================================ > ================================================================ > > > > > ---------- Forwarded message ---------- > From: "ROSE, DANIEL V" <[email protected]> > To: "'Viswa KSP'" <[email protected]> > Cc: "'[email protected]'" <[email protected]> > Bcc: > Date: Tue, 30 May 2017 13:49:36 +0000 > Subject: RE: [onap-discuss] Does ONAP support Multi-tenant Openstack > deployment? > > Sorry I was on my phone so I was recalling from memory. It is called > cloud_config.json and here is how it is built: > > > > https://gerrit.onap.org/r/gitweb?p=mso/mso-config.git;a= > blob;f=templates/default/mso-po-adapter-config/cloud_config.json;h= > 41586c169961da6ec08a6da1008863e3c2f0d07d;hb=refs/heads/release-1.0.0 > > > > It is not exposed as a gui right now. I imagine some enhancement to this > mechanism to allow ease of addition and removal of regions will take place > but all the functionalities you are requesting is there. > > > > Thanks, > > Daniel Rose > > ECOMP / ONAP > > com.att.ecomp > > 732-420-7308 > > > > *From:* Viswa KSP [mailto:[email protected]] > *Sent:* Tuesday, May 30, 2017 6:04 AM > *To:* ROSE, DANIEL V <[email protected]> > *Cc:* [email protected] > *Subject:* Re: [onap-discuss] Does ONAP support Multi-tenant Openstack > deployment? > > > > Hi Daniel, > > > > Thanks for heads-up. I tried to read more details about *lcp* in MSO here > ( https://wiki.onap.org/pages/viewpage.action?pageId=1015834 > <https://urldefense.proofpoint.com/v2/url?u=https-3A__wiki.onap.org_pages_viewpage.action-3FpageId-3D1015834&d=DwMFaQ&c=LFYZ-o9_HUMeMTSQicvjIg&r=2wwdGZ3YcpSivQ2Kio028A&m=punNSpuxdfdye7MHPgaHkL0bUO7SCcsl6CAPrLSmuuE&s=wdR4pFLpJsuZWSrIxj-YGPc_HKWtttDacS_AeAcFjqo&e=> > ) in architecture page, but I couldn't find any. Could you please point me > to right resource? Basically I'm looking for a config page, where I can > add multiple cloud accounts and name it, such that I choose one during > instantiation / deployment. > > > > I would like to try on-board a in-house VNF in ONAP in separate tenant > space. Is there any working group / team working on this item? > > > > BR, > > Viswa > > > > On Tue, May 30, 2017 at 12:07 AM, ROSE, DANIEL V <[email protected]> wrote: > > Onap supports multi tenant deployment as well as multisite. That is the > onap franework can be installed in one tenant or site while the vnfs are > deployed on another. We also have the ability to have regions and/or > tenants for vnf deployement. This is defined by something call lcp in mso > and there is a clouds.yaml that holds all these combinations. The heat > template we put out for the demo does not however have the full > capabilities exposed as its meant to be a simple tutorial > > > On May 29, 2017, at 2:19 PM, Viswa KSP <[email protected]> wrote: > > Hello, > > > > I would like to know if ONAP supports multi-tenant deployment model. If > yes, please point me to configuration guide accordingly. > > > > *More details:* > > > > If you look at other ETSI compliant NFV-Os like OSM, OpenBaton, there is a > specific configuration step, where-in we will have to specify a VIM account > with tenant name, keystone URL. mgmt IP etc. This space will be used by the > orchestra tor during deployment. > > > > While instantiating a NSD, we can dynamically at run-time specify which > VIM account to use. > > > > While I understand that ONAP currently not multi-vim complaint ( only > supported with Openstack ), I would like to know if multi-tenant model is > supported. > > > > Assume I have 2 tenant spaces ( in 2 different openstack sites - to > complicate the problem statement further ) and I would like to instantiate > service 1 in tenant 1 and service 2 in tenant 2, how should I go about in > realising this requirement? > > > > BR, > > Viswa > > _______________________________________________ > onap-discuss mailing list > [email protected] > https://urldefense.proofpoint.com/v2/url?u=https-3A__lists. > onap.org_mailman_listinfo_onap-2Ddiscuss&d=DwICAg&c= > LFYZ-o9_HUMeMTSQicvjIg&r=2wwdGZ3YcpSivQ2Kio028A&m=8aOVDebX_cac4i_ > WaD2VI1ZaKYZWTLkLffNt5I5HUgo&s=MSoC0S256pAGmVajoRWmhow6Wr- > 4B4vNBOr0W_ffXgE&e= > > > >
_______________________________________________ onap-discuss mailing list [email protected] https://lists.onap.org/mailman/listinfo/onap-discuss
