 Okay, let's start then. So we don't have too much empty space at the beginning of the recording. My name is Erdiko VanCe, I'm also working for the OpenStack Foundation and I would like to welcome you all on this community meeting. This is basically a new meeting series and the goal of this series is basically to give an opportunity for projects to give updates and demos on the work that they are doing. This is a completely open forum and we are actively looking for people to contribute and present their work. So if you're interested in being up on stage on any of the upcoming occasions of this meeting then please reach out to my colleague Chris Hodge or basically anyone from the foundation who you have the contact info to and then we will make sure to get you up on the schedule. Another thing that we have is a community newsletter that you can sign up for. I threw the link to the sign up page into the chat so if you're interested please check it out and go and sign up. On the today's call we are introducing one of the new top-level pilot projects of the OpenStack Foundation. It's called Starting Acts. This project is a new software stack basically for edge computing and IoT use cases. It integrates together a lot of well-known open source components from open stack services to Linux, Kubernetes, OBS, DPDK and so forth and it also adds some new services focusing on important things such as high performance, low latency, high availability, everything that's important for edge and IoT. The team is also getting the first release out this week and it is also a growing community so I would like to encourage everyone on the call that if you're interested after hearing more about the project and the project team and the community if you're interested to participate please do so. We have a website for the project StartingActs.io with all the information that you need to know in order to be able to get started also don't be shy to ask questions and as we have two people on the call being way more knowledgeable than I am I would like to give the word to Bruce Johns and Ian Choliff to share more details with you about StartingActs. Okay can we see the slides on the line here? Yes. Okay I'm going to roll through here and hopefully my friend Bruce can join us back in flight here shortly. So my name is Ian Choliff, one of the members of the Starling X community and a member of the technical steering committee of the Starling X project. So as Ildeco was saying we're launching Starling X officially with our first release this week and it's a new top-level pilot project of OpenStack Foundation. We're really excited to be sharing with you today all the great capabilities that this project brings to the table providing high performance low latency and really a small footprint for critical edge applications and solutions today. Certainly looking forward to growing the community we've got a large community already established but lots of interesting technologies for people to contribute to and grow the code base and extend the promise of edge computing and cloud computing in general. So really excited to be here today and share the community aspects and also the technology. So from a project perspective there's a number of technologies and trends in the computing industry today as applications move closer to the edge or require higher performance at the edge whole new requirements are coming forth for things like autonomous driving, IoT in communities, mass transit and really bringing latency high bandwidth to these applications because more and more data is being generated closer to the edge of the network and it's really not practical to backhaul it all but we need low latency high bandwidth at the edge to deliver that solution. In addition we need to build in security because whereas typical cloud solutions are in a physically secure data center many of these edge applications will be in less secure locations at the edge of the network. In terms of the youth cases that this community is exploring VRAN so virtual radio access networks is certainly a very interesting youth case providing the capabilities to serve up high bandwidth low latency to the baseband units as well as the radios so really providing that that linkage and providing a virtualization layer to really radically transform and provide application consolidation at the edge of the network to really enable other use cases like multi-access edge computing where we're bringing virtual reality augmented reality to the table along with the radio network and tying that into things like connected vehicles as well. From an industrials perspective for healthcare we've got some really interesting use cases for imaging and diagnostics how can we improve the patient experience and the clinicians experience in terms of having all the data available to them seamlessly, transparently and being able to have a really informed dialogue with the patient and in a very secure way within a healthcare facility. Some more use cases so if you think about an autonomous car and how that concept applies to some other transportation modalities certainly rail is a very interesting use case where in a freight train there's a huge amount of momentum behind the freight train so being able to adapt the environment around it and provide security for people that are operating in and around the train through camera analytics and really transforming a locomotive into a rolling data center. Same thing in a ship so a ship has got a huge amount of data going through it and being able to leverage and consolidate all of that information and optimize the fuel burn in a ship for example. Also seen applications where we're joining the HMI so the human interface into automation and controls for virtualizing programmable logical controllers as well as providing full digital twins of control systems for simulating and and doing analysis of very complex control systems and really taking that to an edge cloud in an air-gapped environment is critically important. From a micro CPE perspective again bringing content delivery, SD-WAN and all these solutions together where you've perhaps got a hierarchy of clouds with some centralized cloud with again distributed distributed control solutions for providing content delivery at the edge of the network. So what problems is Starling X solving today? Certainly we're seeing a need to drive compute to the edge to do all this processing and minimizing the backhaul. It's really changing a cloud computing architecture challenge which initially started with driving up the scale of computing and providing mobility of applications within a data center into a massively distributed data center that is a really a hugely interesting classic distributed computing architecture that is going to require new approaches and new solutions and that's the thing that excites me about Starling X is how do we attack and overcome some of these really interesting distributed computing architecture challenges. So really the intent of this project is to reconfigure and take the foundational elements of cloud technologies built on top of the outstanding work that the OpenStack community has been doing for for many years now and see how we can take these technologies and distribute them geographically and drive the adoption of edge computing and really bring forward some of these transformational use cases that we've been talking about so far this morning. So we really want to drive down the friction because of adoption of these technologies cloud computing is maturing but still not super easy to adopt and as you drive these technologies to the edge of the network we need to lower the friction make it easy to manage and deploy all these all these great technologies. So let me talk a little bit about the stack that the community has built and where we're going with some of these technologies. So we've built a very scalable and deployment ready solution. Obviously we build on top of a Linux OS bring together technologies like KVM, OpenVSwitch, DPDK, Ceph and others to build a solid compute stack infrastructure. So we build on top of Linux. KVM is the primary hypervisor of this project. We layer in OBS along with DPDK so the OBS DPDK project for virtual switching and then layer in OpenStack services for the cloud control plane. So we're leveraging projects like Nova, Neutron, Keystone, Ironic, Cinder, Glantz, Heat, Swift, Morano and the telemetry family of projects like Panko for example. And last but not least obviously the Horizon GUI is also available as part of this project. There's an additional set of projects that Starling X is managing in the Starling X repos and I'll go into more detail about some of these in the subsequent slides but I'll touch on all of them at this point. So config management is all about managing and deploying the services to the various nodes within the system. Fault management is all about monitoring and recovering faults that may be in this very small edge cloud, being able to provide REST APIs to be able to query what faults are where in the system as well as providing traditional interfaces for some telecom use cases for things like SNMP. Host management allows us to manage all the hosts in these clouds, monitor the services that are going on them and take them in and out of service so that we can actually do some maintenance on them as well. Again as we get to the edge of the network there's less touch available and big barriers to actually rolling a truck to go and manage these remote hosts. Service management is an overall maintenance framework that allows us to monitor the health of every individual software process on the system, asynchronously recover any of those services that may have issues with them and seamlessly recover them and get them back into service without impacting the operation of the cloud. Lastly software management is all about how do you make sure that your cloud is running the latest and greatest software so being able to upgrade from one release of another of this curated stack of projects as well as a framework that allows rapid deployment of fixes to any CVEs that may be in the cloud or bug fixes or bulk changes to the infrastructure within this cloud context as well. So one of the things that Starling X is particularly interesting about is you know traditional open stack clouds are typically deployed in in data centers at very large scale. Since we're focusing on the edge we have to really focus on the on the footprint so we have three main topologies that we're able to deploy a single hyperconverged single server solution where all the compute services the cloud control plane and storage services run on a single node along with VMs. We scale that up into a one-for-one protected pair of servers where this is a critical infrastructure application and the the user wants to be able to be redundant from any hardware level failures as well. So that's the next scaling point. On the far right hand side we have the multiple server configuration where there's a top of rack switch dedicated controllers for the open stack services and Starling X services for the cloud control plane dedicated storage nodes and this really allows much much more scalability and a variation of this is also where we would have a centralized cloud control plane and remote hyperconverged nodes where you're able to have manage those across a geographically distributed set of sites as well so but it all provides a single plane of glass to the cloud operator. So drilling down into some of the services at the next level of detail as I touched on configuration management is about managing and installing the initial cluster but also making it really easy to add new nodes to the cluster and manage all the all the configuration through a declarative framework. It also allows us to provide new nodes that are added to the cluster with personalities so it's a new node, a storage node, a compute node or some other function. As we talk about higher and higher performance solutions the inventory discovery allows us to configure and provide visibility into the huge page configuration for memory configuration for very high performance applications as well also making things visible like GPUs and cryptographic accelerators as well. On the host management side we're able to manage and deliver the full life cycle management of the host so bring a host online but then bring it in service and then be also able to monitor if it does fail and then use that information to migrate VMs to active nodes and things like that so very critical during the management phase and the upgrading phase of the cluster as well. Host management also provides a framework where board management BMCs are available to do out of band activities like power on off also provide visibility into the cluster in terms of any hardware sensors thermal fan again gives a next layer down of visibility into the infrastructure and allow recovery and maintenance actions to take place and all through an easy to use REST API. Software management is a great framework that allows people to deploy corrective content or security vulnerability fixes out to out to these clouds be able to manage and automated use an automated framework to roll through a full cluster so you're able to add a patch to the controllers and then through a programmatic interface the system will automatically upgrade all the hosts migrate VMs around the cluster to minimize the impact on the running application so it looks at the available resources and uses live migration to seemingly migrate seamlessly migrate these workloads around again we use the same framework to do upgrades from one well we will do that in future future releases of Starling X where we'll be able to migrate from one release to another and be able to do this in a seamless way so that the framework can be leveraged to do this in an automated way so you're not tying up operators with a lot of keyboard work to migrate through all the hosts and get them up and running to a new release of Starling X so very interesting framework that allows people to do some some pretty cool things in terms of making sure the the security of the the cluster is up to date and operating optimally so that that's a bit about some of the the frameworks and technologies we're leveraging today what Starling X brings to the table in terms of additional services and how it all fits together into an easy to use and seamless framework for for the edge I'd like to move on and talk a little bit about where we're taking the technology and enabling some really interesting new and exciting technologies that are around in the industry so we're really turning the stack on its head in some ways the community is working on today containerizing all the open stack services to make it even easier to deploy and manage with a zero touch provisioning framework along with a declarative framework leveraging projects like OpenStack Helm obviously having a local Docker registry and running Kubernetes services to manage all the all the containers as well the initially the Starling X services will run on bare metal but our our view is the architecture will evolve and those services will also migrate to containers with service management which is monitoring the underlying all the underlying software processes you know we'll have to figure out some interesting ways to manage that in a container so again some really interesting problems as we move to a more of a tainer centric environment and really allows us to do a lot of flexibility in terms of quickly adding new services so if a user wanted to deploy and configure a new service they'd be able to update a declarative framework and then deploy that in a very cloud native kind of way and you know we're really taking these concepts to the edge of the network again focusing on making sure we have a very small footprint for the solution and making sure that this fits in one and two node configurations and then scales up into a distributed edge cloud configuration we're using Calico as the initial networking plugin bringing stuff to the forefront and leveraging Helm and other other projects here so we're really broadening out of the ecosystem this will provide you know continued full support for virtual machines and building on on the container support that's present in in the current release for applications but also taking it to the next step where we're containerizing all the Starling X and OpenStack services in the future so a really interesting and exciting part of Starling X and really encourage people to join us and bring their ideas to this exciting next generation technology so so this really helps accelerate the adoption of edge technology we've got a lot of building blocks out there that people can make choices on Starling X helps accelerate that by providing some choices and building together a full stack solution that people can deploy that's been tested and ready to use for people today so I'm going to pause here and see if Bruce has been able to join us or I will I am here can you hear all right I can hear you Bruce over to you my friend computers are such fun so next slide please so we're very very proud and happy to be part of the OpenStack Foundation and part of the OpenStack community and we're fully committed to the four opens of software development we look for our technical contributors to make all of the technical decisions under the guidance of our technical steering committee which I'll talk about in a minute we're committed to building a diverse open and strong community next slide please so this is actually a fairly large project and in order to help us manage the complexity and to you know kind of reduce the scope of what people have to think about as they're working on it we've divided the project up into a number of sub projects each of the services that Ian has been describing for you today as its own project they are slightly dependent on each other right now and we're working to reduce that dependency in support of those main projects we have a number of other sub projects for documentation for tests for release for security we also as part of the build we include a number of open source packages and so we have sub projects that help us manage those life cycles we're constantly scanning for updates from the upstream packages for fixes and cvs and etc each of the sub project teams has a lead a team lead a project lead they will have a number of core reviewers and of course contributors and I'll talk about all of those on the next slide so we've borrowed a lot of our governance from both the open stack community and the kata containers community who've done some really pioneering work and really good work and defining how to run a well-run open source project so we're calling a contributor someone who's made a contribution in the past 12 months and anyone who is a contributor can run for and vote for the elected positions that includes the tsc and all of the leadership roles that I'll talk about in a minute a contribution can be code it can be test it can be documentation um we're trying to keep this simple and automated so we really look in the garret cues and see who's contributed and then if needed the tsc can provide contributor status for people who've made other contributions to the project that don't necessarily show up in garret we're also using the concept of a core reviewer these are active contributors and experts in a particular technology area they're appointed by their fellow core reviewers their main responsibility is for reviewing changes and specifications and the cores have the ability to merge changes into the garret tree the next slide please so we've we've split the traditional open stack ptl role into two separate roles in starling x so we have a technical lead per sub project that sets the technical direction for that particular sub project we also have a project lead who's doing the coordination work and the tracking and the communication and serves as kind of an outside facing ambassador to the project and we're doing this for a couple of reasons one is that we've been blessed with the people on the on the teams who can fill these roles and second reason is is that there's a specialty of skills so people that are good at project level work aren't necessarily strong technical people people that are the strong technical people aren't always good at the project level work so it's a way for us to optimize the team and to let people work on what where their skills are best suited next slide please so overall the project is managed and led by our technical steering committee they're responsible for the architecture of the project for approving new sub projects for archiving sub projects that are no longer needed or no longer active it's also the place where the buck stops and any final decisions or escalations or disagreements with the team can be resolved in the technical steering committee the current governments governance says that this will be seven people the initial group will be appointed that we will move to an election-based system within the first year of the project the initial members are Brent Ian Dean and Saul let's go on so there's lots of places you can start looking at our code is available in git garret we have a wiki that contains informal documentation we're in the process of moving that documentation into the the a document documentation repository where it can be web accessible if you go to starlingx.io today you can start seeing the results of that I'm very very pleased with the work that our documentation team has done in moving the content off of the wiki and into the formal track documents we're tracking our bugs in launch pad we're tracking our stories and storyboard and we're we have a specs repository where you can see the currently active and approved specifications that the team is working on next slide please so you do not need to be a member of the open stack foundation in order to contribute but if you'd like a vote in the foundation board director's election you probably would want to join I strongly encourage people to join the foundation it's a pretty cool place if you're contributing on behalf of an employer then please make sure your employer has signed a corporate contributor license agreement next please so we have a free no channel we have mailing lists we have email we have weekly calls we actually each of the sub teams has a weekly call the overall project has a weekly call the tsc has a weekly call we encourage everyone on this webinar anyone who may watch it later to feel free to join our mailing list so check us out on irc welcome to have you attend the weekly calls and thank you very much we're happy to take any questions that you have thanks Ian and Bruce I see questions in in the chat so I will start with reading in those and then if anyone else has questions then either type it up on the chat or just ask it after these two are answered so the first one is from Andrew he's saying footprint consideration is key as mentioned what the actual memory requirements for a single server deployment and what are the system CPU needs so the the memory footprint would be dependent on the number of VMs that somebody would want to deploy so it would be not just the platform infrastructure but you'd have to take into account the the amount of memory depending on the number of VMs that people would want to run we have some recommendations in terms of anywhere from 32 gig to 64 is typically what we what we see in terms of cpu footprint we can fit the base services in a one and a two node configuration into into two physical cores so very very small footprint okay I guess the question was obviously the number of application of the determines the amount of memory and cpu for applications clearly but the question was more about the system overhead right so on the cpu side it's one to two cores okay and the memory again it depends on the number of VMs that you want to manage and what services you're running but you know sort of guidance we give is 32 gig okay thank you okay the next two questions are from Phil he's asking about the first target use case you are looking for so there's some natural use cases for Starling X there's a number of telco use cases if you think about putting edge servers at the base of cell phone towers and be able to manage those remotely those kinds of use cases are fairly natural we also think that there's a number of use cases around factory automation and industrial deployments where moving computing resources closer to the manufacturing line and doing more processing closer to the manufacturing line can benefit the the overall manufacturing process thanks the the next question is again from Phil he says I understand the technical details are based on open stack and containers and there are similar projects to this and he's asking about the differences with other mac projects so I'm probably not the best person to talk about other mac projects I'm not sure what what projects he means Phil are you there can you maybe clarify the question so I'll go ahead and try to answer anyways I think the the the value proposition here is in a couple areas one is the the new unique services that Starling X provides that simplify the manageability and the upgrade ability and the overall maintenance of the platform and the second is is that the system is validated and tested as an entire stack so the components have been optimized and tuned and validated together to ensure that they all work and can be easily deployed so should we move on to the question from mark about the integration with Starling X and Nova yep so I'll I'll take a shot at that and Ian may jump in there are a number of changes out of tree changes to Nova in Starling X we are actively engaged with the Nova community to get those upstreamed get them accepted by the Nova community or re-engineered in Starling X to align with Nova we're actually at the PTG we had really good sessions with Nova with Neutron with several other open stack related projects the Starling X team is working very hard to reduce the patch backlog that we are carrying right now it's very important for us it's very important for the community we're down by a third of the open stack related patches since we started the project so we're making ongoing and continued progress to address the the delta between Starling X and the upstream masters yeah and I would add at the PTG in Denver there's a great dialogue with the Nova Nova community you know there are a number of specs that we've contributed to the Nova project and they're under under review by the Nova team right now and you know that that will continue the collaboration with Nova team and you know looking to accelerate those reviews cycles so there are people on the call to who are interested in that please please get on and promote the reviews that that you see there happy to happy to get input from folks sorry if I could add follow on to that do we have an idea of a timeline for what release these might be landing in well I think that depends on the adoption of these other projects so we've got to work into their existing work work queues and and backlog so you know there's there's a lot of review cycles that go on so it's something that we're actively working release over release so there's there's specs and changes submitted to Stein already across several of the open stack projects and everything we want won't go into Stein so it'll go into whatever comes after Stein excellent thank you so who do we contact to create a proof of concept you can contact the community on our mailing list you can reach out to me directly you can reach out to Ian directly either one of us would be more than happy to help with any proof of concepts of Starling X Ian can you talk about the GPU question I don't have the details on that yeah so the the approach right now is is to leverage the GPU pass-through capability and also work with vendors in an open way to have a V GPU so virtual GPU capability and uh those vendors are working on some of those technologies so it's really a vendor vendor agnostic approach I am aware of community members who have done done work with Nvidia and others and like Intel for example as well so I'll take the acreno question um what is the relationship in roadmap vis-a-vis acreno and Starling X so um both uh both Intel and Windriver have representatives on the acreno technical steering committee we're very much a part of the process I'm part of the acreno community myself there are blueprints that are being submitted for Starling X within the acreno community there's work within both communities to try to align in general the the plans that Ian discussed for containerization include using airship which is part of the acreno stack that AT&T is defining in their blueprints so I think the answer to the question the short answer is the community's overlap and there's several of us that are part of both yeah I'm also a member of the acreno community and and you know it's a community that's evolving and we we plan on that continue our our work in that community and making sure there is good alignment between Starling X and acreno okay so the question on the state of OS abstraction is something we're just starting to work on it's a very large effort the software updating software upgrades that Ian described right now need abstraction layers built into them in order to handle other package file formats validating on a new OS maintaining additional OS every OS we add to the project multiplies the validation effort so we're looking at but we're very much looking at moving in this direction we have a sub project created called the multi OS sub project for addressing those issues and working on them and it's all being actively worked in the community right now yeah I think it's really important to highlight that the multi OS project is just getting started and you know providing that layer of abstraction that Bruce mentioned is is what that sub team is focused on and you know we're looking for expertise and contributions from other other groups and other Linux variants to help this multi OS project provide that abstraction so that we can provide that base OS flexibility to starling X users and developers we're also working hard to make the services more independent so they can be used in other ways and other systems there's a lot of value in the technology that's that's been open sourced here and we're hoping to see that value fully realized other questions it seems that all the questions are answered so with that I would like to thank you again Bruce and Ian for the amazing presentation for those of you who are interested there are a lot of links in the zoom chat including the the website starting axe.io the website contains all the links that you would possibly want to have like to the code the documentation communication channels such as mailing list and IRC channel so if you have further questions please reach out to the community and ask them on the mailing list or on IRC this meeting is also recorded so we will post the recording to the website as well once it's available so you will be able to listen to dad and I also mentioned that there is a community newsletter that you can sign up for the link for that is also in the chat as well and with that I would like to thank you all for participating and see you all on the next webinar slash community meeting and everybody thanks this is Chris Hodge from the open stack foundation and I want to thank everybody for joining us this is kind of our second big community meeting for the first few of these we're actually doing them at a very high level the previous week's meeting was about project governance if you follow the link to the ether pad in there you can you can see the the slides to that to that meeting the first few of these meetings are probably going to be a little bit higher level but as we move on I really want to encourage everybody within the entire open stack foundation community across all of the strategic areas to contribute you know talk about exciting things that you're working on and you know really make this this meeting about you know kind of showing you know sharing with the entire community the work that we're doing so thank you for everybody who came to this meeting and if you would like to contribute to a future meeting please feel free to contact me at chrisatopenstack.org thanks and with that I think we'll we'll call it into the meeting