 I'm I'm the lead systems engineer for IOT and embedded compute for EMEA within Dell technologies Dell technologies is consisting of these seven companies that you see here The talk this morning is about a Jax foundry See the octopus there I'm gonna explain a little bit more about what the octopus is all about I Jax foundry is part of the LF edge Linux foundation edge initiative So I've got 25 years of embedded experience and recently the one-half year joint Dell technologies So let me go over the agenda Quick history Some backgrounds on the Ajax where it's coming from we're going The architecture and the technologies used inside the Ajax Foundry the current status How the ecosystem is set up some of the governance that's used and some of the releases I'm gonna dive into some of the road map Got some people here that that are actively working on on this as well. So Some use cases and some of the hackathons that we've been having this year Also, we're planning on doing next year and Some new initiatives which have been recently announced this Monday Going over project Alvarian and I'm gonna explain what what that's all about. So Why is IOT difficult? Basically it is difficult because it's it's the The post-doctorate of everything that we've been doing for the last 35 or 40 years in in I.T. You have network protocols mobile computing cloud computes AI machine learning You need to know it all that's that's why it's hard and how do you choose because You need domain expertise in analytics data usage security But also how do you handle the connectivity soup of all these different industrial protocols that are out there? Then there's the application environments. I'm gonna use see C++ Python.net How about operating systems? So you need to have a lot of choices and Essentially what you want is freedom there to choose whatever you want, but still have Have support in in your ecosystem that you're working on Now Ajax is is the answer to that. It's it's open-source fan and neutral projects. It's more like an ecosystem of Work It's based on microservices and a loosely coupled framework for IOT and edge computing It's hardware and always agnostic And it's a Linux foundation Apache 2 project. So basically it helps to to yeah Have a common building blocks and APIs to build On an IOT platform rather than having a siloed IOT platform Which is not adaptable in a greater picture. So if you want to do this at scale you need to To look at Ajax Foundry So the overall goals is to unify edge computing and have an open platform for people to add value into that platform Have an ecosystem where everybody creates plug-and-play components That can be certified in order to work together in this ecosystem and provide tools To to actually faster get your products to market And work together on all of these open-source projects and ultimately enable and encourage the growth of IOT solutions So a brief history of where we're coming from it it all started in 2015 where it started as a CTO incubation project project views Done by my CTO Jason Shepard where I report into Then invested seven many years in that initial project and gave about 125,000 lines of codes As a start of this this project Was mainly sponsored also by Jim White who was sort of the brains behind all of this It ended up in open source in April 2017 with 50 members and it now grew up to about 76 members And it's part of the LF edge since January 2019 So we have a cadence of of two releases a year at least that's what we're trying to do And we've been pretty successful up till now So here you see the the actual framework in the flash, right in the bottom side you see all of the Device services basically you're interfacing to the outside world So we have good set of of protocols and we have actually have a lot of people Creating their own Device drivers for instance. I'm working with a sensor manufacturing in Germany that creates their Device drivers for all their sensors that they have in a portfolio So that's that's really good because that that gives a lot of stickiness to this project Yeah More people could do that that that would be great So then we have sort of a core services in there where all of these information is traveling through Then we have some support services on top of that With your scheduling you rules engine basically where you do some of the analytics as well And then you have your clients and application export services, which is basically your access to the cloud So on top there there's there's quite a lot of changes I'm gonna go over that later on So on the on the right side you see your management layer Which is basically helping to support all of this and make sure that everything that's in your data Plane is is still working on the other side You've got your security bits, and that's that's making sure that that's all of these can be done in a secure way So how does that that work basically you can read a temperature over back net That will be landed into core data that core data will then go into the distribution It will go into your on-prem or enterprise or clouds Infrastructure some analysis on there and basically it travels back through your rules engine It can do some local analysis on this Before it actually goes into the command and that command can go into the MQTT driver And basically say okay, we'll stop the machine right there because your temperature is over over a certain threshold So Since these are all microservices, there's a lot of Changes that you can do for instance if you need to real-time capabilities or you need anything else You can do that on the fly and basically get that into your platform Also, this can be distributed, and I'm gonna show you how a little bit later So This is sort of the biggest changes that we've gone over in the the Fuji release which has been released actually Monday So what we're gonna do with there is replace the top-level services with an SDK type setup Where whether you talking to Azure Google? AWS or anything else you will be able to do that And you can actually write your own export services there if you wanted to So here you see some of the tier deployments that that are actually Real and we're doing this with with customers at the moment So on the far left side, you see your field devices which get all the sensor data Over back net and Zigbee So basically you can have some some rules there to measure some temperature and lightning settings There it can also take some some other pump and camera feeds and feed that into floor-level Gateway, which then can do some of the higher-level analytics So that can be communicate to to a building level device before it actually goes into your your Cloud analytics where it can do some deep learning and other mechanisms So really distributed setup and you see that whatever you need inside the Gateway at that level you can actually house that but it could also be sensors that actually Incorporate some of these services already in these devices, right? So depending what you need it's sort of Distributed also you can divert in two different directions to do some local analytics on a data center While you're making decisions at a lower level So some of the performance targets that we that we look is for instance a raspberry pi 3 1 gigabyte of RAM 64 bits CPU with 32 gigabytes of storage space It should start up within about 10 seconds depending on the database that you're using And the typical latency between the ingest and the actuation should be less than a second You see here also the Dell Gateway and some of its metrics and how that performs in into a typical setup so Some of the Architectural tenets that that we have so it should be platform agnostic, right any distribution or OS protocols and sensors. It should be able to to work there. It should be extremely flexible It should have these store forward capabilities for disconnected devices And it must support facilities to moving closer to the edge So everything your decision-making should be actually happening as close as to the sensor as possible Before it actually gets transferred to late to an upper level Due to latency concerns or bandwidth and other concerns Should support green and brown device sensors And it should be secure and easily management manageable So here's some of the open-source technology that are actually used inside Ajax family so We've refactored. I think in an early stage everything from Java to to go Some of the device services are in C and C plus plus there's Javascript involved We can basically work with any language there Rest API to communicate between the different services So monger DB or Redis is a choice that you can make there So also the the message queuing can be can be different And we use a lot of other open-source technologies like console for the configuration and registry Kong for proxy Drills for the rules engine. So a lot of a lot of technologies open-source technologies that are built into this project now basically what it helps you is to prevent you from locking in to to any of the existing cloud Providers because they they try to lock you in and once you're in you won't be able to get out because you've basically put all your effort in there That's what you want to prevent It needs to enable and secure a manageable solution Basically, you can certify components working hard to get that that's to do a more mature setup It helps you to to Get all that protocol soup sort of manageable in a good sense and Because it's part of the LF edge Initiative it's it's it's very open and it can interact with a lot of other ecosystem partners And it helps you to digital transformation In that sense that you can do a multi-cloud solution with this product as You prevent that that lock in there so as I said, there's a lot of Traction since 2000 to the start of 2019 where we basically became part of LF edge and With that it helps to have sort of a focus on everything that's happening in the in the edge Which was previously IOT Delgate cut away from that So What you see is that the total marketing capital which which is in there is about 300 trillion Dollar and the funding in total of this project is close to one trillion in total so it's it's heavily funded The whole LF edge set up So here are some of the the members that are part of the LF edge Ecosystem you see all the big names in there And it's it's a good foundation to actually get this the support from that that ecosystem Same goes for the L at the Ajax Foundry Community, which is You see here There's a lot of big names that that are in that ecosystem and all support the growing ecosystem of Ajax Foundry Now how successful is that? It's it's it's really successful actually if you see the the rise of contributions and the downloads of For instance the Docker containers that that are associated with this you see a huge Increase happening Actually this Monday, we reached one million downloads, which is which is insane. It's it's going really fast Getting really excited about about all of this so you see a lot of traction so If you want to engage Basically, it's a it's a meritocracy Anyone can can contribute and and be part of this this ecosystem There's a technical steering committee and a working group I'm going over a little bit of that in my next slide So there's chairs and in there if you want to contribute here's some of the GitHub repositories where you can go actually download the code and By basically that command that the W get command you can start directly Testing it and check it and see if your APIs are running I have a video which which shows how to do that. It's it's literally Minutes to to set you up and have something and have some data running Also, we have a snap from Some of the people at Canonical that actually supported us in this Which is a bliss to actually start start installing that so if you want to contribute or are getting gates with that I Will make these slides available I haven't been able to do that because there was some confidential information in that because of some of the Announcements that we're making but I will make these available for sure to you So how is the steering committee set up? So basically it's it's it's governed by by Jim white and Keith steel Both of them currently working at Io tech. Fortunately, Jim white left us at Dell and he's going to Work for Io tech which doesn't mean anything for the project because he's still gonna be in this this ecosystem So from the Dell side, we have Trevor Khan who's part of the core working group So we have people from VMWare people from Intel all in this this ecosystem So it's it's it's a pretty good good setup that we have on in terms of the governments It's it's getting bigger and bigger. So I won't be able to attend any of these sessions anymore Because it's it's just to getting too big So in June, we released the 1.0 version, which was the Edinburgh version. It was a basically a result of the All the work that has been done in in Edinburgh in the UK in 2018 So improve the onboarding I'm gonna have some some different binary data for support Automate some of the performance testing that we needed to do because that that was difficult We added many device services to the set up And the application service was was more scalable, but as you know, we're gonna change that now so we refactor all the database setup and The certification program was was outlined. So it wasn't completed outlined so this week we released the the Fuji release and Basically, it had all the Ingredients in there to to make it more performance. Also make it measurable in the in performance Improve some of the security and include better PKI management in the services The store and forward capabilities So a lot of work also on hardening the services and the SDK because that was needed and the the Ajax marketing work group, which was Basically lost because the LF edge initiative was formed. So that was picked up again So for 2020 what we're targeting is actually getting the third party certification Completed that Is important to to actually get get that that X mark on to all of these different components from various people Work on some of the the setup guidance and some of the performance measures and What we're gonna address is also one of the Initial us that was asked to me last year in in Annenberg is high availability concerns specifically in Getting getting East and West Support for for this project is it's one of the biggest asks. So we're gonna attach that in in the 2020 release So some of the use cases that that are actually happening you see here So we have dev kits and the physical architecture you see here With a recipe pie and some of the growth material So what was done here is actually incorporate MQTT broker and and no red to actually dashboard all of this, but also do some of the logic here to get it onto a web browser and Actually manage what some of the sensors that were associated with this if you want to Have a look at how that's done the link is under there And you see how that fits into the small picture there We're basically the analytics part is being exchanged by by notepad very cool setup So With all of these technologies and open source initiatives What we see is that there was an ass to actually Incorporate more stuff into this and that was our internal project Magnolia Where we wanted to have also the RSA agent and and some of the Redis code included into the whole architecture and Have a different OS on there, which was the photon OS from VMware so With the people from Io tech We now have Commercial support for for these types of setups and we're being pretty successful in in larger POC's With specifically in a Maya with with customers That that actually are doing works with multi-clouds and getting all of these device services in there So if you want to do this at scale, yeah, you see that that you need professional support on that If you're a smaller company But also if you if you want to use all of these technologies together can be somewhat overwhelming So Io tech can can help there As well as VMware, which is also within the photo. No as shipping adjects foundry at the moment So that that's pretty good good initiatives So in October This year We had a hackathon in in Chicago and that was a big success Basically it was around Some of the retail initiatives that are currently going on some of the use cases were advanced loss protection Some of the personalized retail experience that that you can do But also inventory management for instance with handhelds and drones And there was an open category with with which needed to have that same retail centric use cases So we had four teams in there. I wasn't present So I needed to get most of this from the media. I haven't been able to speak to to anyone there yet the team from Voltaio one that they got a check from 5000 euros And you see here that they got they got that check handed off their two days of hard work And bringing that to to success what they did is they used camera in jest to actually get to get that whole setup initialized We're planning to do a Manufacturing Hackathon in EMEA in the spring of 2020 That's gonna be right before some of the the new releases of of Ajax And it's it's gonna be probably around to Hanover mess so if you're want to Compete with with other contestants there that might be a good good way of doing that The location is not yet set. It might be in France. It might be in Germany It might even be somewhere completely different, but we're still looking at that at the moment so some of the other exciting news this week was project alvarium and project alvarium is basically and you've seen that in the presentations in the keynotes this this week The concern of where's my data coming from and how trustworthy is that that data? So project alvarium is actually addressing that need for the industrial space There's a session later on today, and I'm going over what what it's all about But basically what what we've created is a proof concept for a data-confident fabric And the data-confident fabric will help you to get it with a little bit of blockchain and a little bit of TPMs and all of that to formalize a Setup which basically allows you with open source technology to bring all of that goodness on trust Into a score mechanism where you can basically see okay if it's gone through all of these systems Then how trustworthy is this data when I'm actually looking at it because that's becoming increasingly important You saw the picture right off of the the the mouse actually with some noise can enter into for AI a monkey now with this you could prevent that and It will do some some things which for instance for digital cinema is there for years But now it actually becomes Relevant also in an industrial space, which which we look at here so it's a collection of Various trust in search and technologies that actually bring this all to to open source So it looks at the wider system and it's it's It uses all these different technologies under the LF edge umbrella so it's it's actually nothing new, but it brings all these technologies together and Makes that the proof of concept and it's sort of a working method on how you would do things so here's some of the initiatives the open source trust in search and technologies and how that actually will equate into Bringing all of this together and it's It's actually really cool to see also at six foundry being part of that that whole set up so Why is it in LF edge the reason for this is that all create all data is created at the edge So that's where you need to to actually have that that sourced and You see all of these other Incentives coming into that that technology CNCF elephant networking So it's it's unifying it it's not reinventing So if you want to know more about this There's some other sessions later on today You see that traffic con One of senior staff members and director for for Dell technologies Will have a session at 415 On how did it all being done? and It's it's pretty impressive. We we used a lot of Dell technologies VMware technologies Boomi to actually get it to get this done and yeah so also some of the other team members from for instance from VMware you see here and Yeah, I encourage you to to look at these other sessions Later on today Thank you very much