 Welcome to our presentation day three operations new region and new services in this talk we would like to highlight a little bit our view on how to organize the future of the public cloud and how we at the open telecom cloud see that. My name is needs Magnus and together with my colleague, Sebastian, I'd like to give you this short overview. My name is needs Magnus I'm an architect for the open telecom cloud and I'm also acting as community outreach manager. And well, with me is Sebastian would you like to introduce yourself as well. Thank you, my name is Sebastian winner I'm the chapter lead for the engineering of open telecom cloud. I'm doing Linux and open source since 20 years and more and yeah I'm with open telecom cloud really since the beginning and started this whole endeavor which we would like to introduce to you and then talk about how we do operations and how that evolved over time. Yeah, could you outline a little bit how we all started our collaboration with open stack. Yes, of course. So, as said I'm with that since the beginning. We started in 2015 to build up a cloud based on open stack. The target was to to bring a public cloud to the market in European governance in German governance, obeying data protection and privacy laws. But also on the other hand side being open and giving the power of open standards to the end users like open stack is doing that. So we started in 2015. And finally I dig out from the archives, a PowerPoint slide with our initial start of services what we had or set of services. As you can see here really infrastructure as a service with a few additions. But if we look at what we have today it is really evolved a lot to what is now available on our cloud. From that beginning between 2015 and 2016 when we launched open telecom cloud and see it until now and then this is from May this year, you can see a lot of things have changed and then things evolved from Yeah, I said infrastructure as a service with a few things to a whole set of portfolio. What we do bare metal machines that we manage containers that we manage on the other end of the scale. The databases big data upcoming artificial intelligence really ready made functions to to just use and then leverage for your business that you want to run. But to do that also not only did the cloud and the technology had to evolve but also we as a team had to evolve and this is what is widely discussed throughout the last year as they to operation so also having an agile team and and whatever is needed to not only set up a cloud but also run a cloud. And to give you an idea, not only in terms of service what we are doing but also in terms of size what we are doing some some facts and figures here around open telecom cloud. So you can see we scale this thing beyond the normal limit so 500 nodes for a typical neutron installation. Well, we did that by factor 10 and more. So these six and a half thousand servers this is pure compute power available to to the end users, nearly half a million of vcpus and more than three petabyte of Ram. This is virtual machines from small to to big object storage block storage additional services as mentioned before connectivity really in 300 gig and also equinix cloud exchange where we are available, but also as as really a unique thing in the market. We also have a hybrid cloud so if you do not want to run this as a public cloud or you want to have a mixed setup. We can put a hybrid cloud to your premise installed that they are in T shirt sizes from small to large, whatever is needed we we can offer that for you. As I said to run that really we we needed to make changes to how we work and what we work and follow the the agile principles to to get that all installed and ready. This is, as we said, day two, but we are talking here about day three operations and day three is then how to make the next step, not only to have a single side one cloud, but really how to grow further how to adapt to these challenges how to move your technology also forward to be ready for 2020 and whatever is coming beyond. So how to grow that that was one of the topics also for us, as you could see in the numbers before we really scaled it beyond the usual limits but is the way just to to grow further in that limit. Because is no just yet another availability zone there is not the answer to to everything. So it had to be a new region and we were looking into Frankfurt we were looking into Amsterdam I said public cloud what do you need internet, where do you find internet is Frankfurt or it's Amsterdam the two largest cloud or not cloud internet exchange pops in the world. So, after a bit of back and forth we decided for Amsterdam. And yes, climate is changing waters rising we are well aware of that and we picked a data center provider which is also aware of that and is well prepared. Nevertheless, we are doing everything also to do our part of the job to protect our climate and then not to use too much energy. I will come back to that in a minute. But growing is not easy. It's not just as you can see here, putting a few servers in a wreck. Sometimes you are surprised by some challenges which you do not think about. For example, SFPs as you can see here we deployed like 12,000 SFPs. One of the problems was they were all coming in a separate package so it was just boxes and boxes and boxes and even more boxes. So we had to unpack like 12,000 SFPs. What we installed in Amsterdam as you can see here is around about 2000 devices, 1400 servers 500 network devices and all kinds of other stuff. A few more pictures also we had some stacking problems, even they were not related to open stack. As you can see here also logistics company thought it was a good idea to stack servers on servers turned out to be not a good idea. Yeah, and also for the tech guys amongst us this is my let my engineering hard beat faster to see a bit of cable porn here. So clean and nicely this is all installed. And this is really one big part of the story. You need partners which you can rely on because what happened. 2020. Well, nobody thought about that and everybody was complaining that the game of Thrones finale was the worst thing. Well, we had 2020 and we have Corona and COVID-19. So we installed a data center if you can't go there. That was faster question. So the answer, not the chicken or the egg was first really the raspberry was first and to give you also a bit of technical insight how we tackle modern modern problems. We had a data center. And as you could see before we had a lot of switches and servers and cables and everything that was installed. And we had a good partner to make that according to our plans but how do you then do the next step if you're not allowed to travel and you're you can't go there and the data center folks won't do the installation for you. So luckily we had some Wi-Fi which is connected to the Internet. We set up a virtual machine on our existing cloud and put a open VPN on it. And we sent a pre configured raspberry there which was able to connect to the to the Wi-Fi and through the Wi-Fi made an open VPN connection dialing home to our virtual machine that was running there. So we had out of band access real out of band access so we could from our home from our office access the raspberry which was sitting there. We pre loaded it with a repository so it had all the software firmware whatever was needed to do the first provisionings. We installed software for zero touch provisioning to do the network stuff but also some pixie stuff to to provision the servers. And with that one, we are sitting there as said in your office, we could make the connection through the VM through the Internet through the Wi-Fi to the raspberry and then start provisioning so we really could do the first switch painted in magenta. But also the first servers and then one by one we worked our way through the infrastructure. So that we could install software deploy the first operating system and then really have modern provisioning tools at hand to then layer by layer deploy our cloud in the data center where nobody of us was able to go. So this was really a challenge to achieve this but by this simple solution it really worked and as a side benefit. Also we put a serial cable or a USB to serial cable on it. We had a mobile serial console for all these devices. Yes. What else. What did we learn a lot. So many many small things matter like the color of light. So SFP is not SFP even we had so many of them cables you need enough of them more than you think communication to stay in touch with all your people with also the data center folks and their willingness to to use something like WhatsApp just to check a cable on SFP really helped to stay ahead of the installation. Also on the other hand side excavators seems like summer is excavator season we lost so many fiber cables. Really but also having backup solutions in place on redundancy turned out to be very necessary. I'm here I want to mention I said before we also try to do our share. We are not only using x86 but also arm they are much more energy efficient. And just to deploy some standard management notes whatever they are perfectly fine so we we really deployed large scale arm hardware. And they do the job really great to to run our management zone and then everything else where we can use it. Nevertheless, also, we were talking about day three operation. It's also the tooling and modern solution that are important. We had triple round for quite a while we have open stack based installations for quite a while container based open stack installations for quite a while. But putting that all together really is also a large part that we adapt to the technical challenges to have seamless upgrades and then deployment methods and then really a lot of things that we needed to adapt from just hand deployments automated deployments to the way or the point where we are today to run a cloud of that scale. At that point I would hand over to needs to really talk a bit about the surroundings and then also the implications of of doing day three operations. Thank you, Sebastian. Indeed, setting up and managing our infrastructure for scale for growth is one thing. We have also a lot of services on top of this infrastructure and this these services are ever growing and evolving. And if we look a little bit in the past of information technology, we see one pattern that reduces step by step the details of your it setups and at the same time, the abstraction layer increases or the amount of abstraction increases. So, when you in the past, often configured specific snowflakes that are unique, unique setups. We see more and more pre configured pre defined and bundled solutions that come shipped readily to use. And to explain this, I made up a small example. Think of DBA 10 years ago. So when you had an IT project, which was focused on setting up a new database server, and you walked over to your DBA guy. He probably asked you a lot of questions. So what kind of hardware would you like to deploy for that. What drive technology should we pitch is our spinning disk like as sufficient or do we need SSDs. Do we have a storage network dedicated to that or do we need to set up one. So what kind of database engine do we need. We want to use my secret Postgres Oracle Maria DB and so on and so on. There are back ends in terms of the different databases. And if you remember the configuration files for for databases there are long files with lots of options and buffer sizes and configuration options. Okay, this is different. So DBA in 2020 just issues something like an OpenStack database cluster create or something like that. So here in OpenStack we have a trove for example, in the OpenTelecom cloud we have some additional database services as well, but the idea is the same. So today we have pre configured setups that fit many needs and still can be configured if necessary indeed. But the overall time to market is much decreased and it's way easier also for simpler setups to bring a service like a database cluster configuration manager or some other components into into place. Could you turn off to the next slide. Of course. What I've prepared here is the CNCF landscape and the CNCF landscape lists more or less all projects that work together or help support setting up Kubernetes cluster and Kubernetes framework overall. And that leads to two interesting questions. First is where do these setups run. And we've seen often the idea that, okay, let's define something like Kubernetes or some other container management platform as a management platform for for any IT workload. But the question still remains where do these clusters run upon. So what is the the infrastructure layer. Of course you could just use say a bare metal service for that. But even those bare metal servers need to be managed in some way. And especially if you are talking about scale, you always come back to all the nitty gritty details that Sebastian explained to us a few minutes ago. So summarizing this all day free operations mean means making a balanced and informed choice and to summarize the principle of day free operations. We learned that it's starting with the day one operations and configuration tasks as we've seen setting up everything that is often the fun part. The two operations are made about maintaining and keeping everything up and running. But we still have to think about growth and adapting to new technologies about a different cloud scenarios about innovation. How does it fit to business and so on optimization and so on. And that is what we call day free operations. And this is something that we take great care about. And that's already that already concludes our presentation on this strategy. Let me very briefly point out to our open stack scavenger hunt as service that we are a small riddle that we offer here in the light of 10 years of open stack. You have to search a few websites and then you can win a cool gadget price in a raffle. So if you're interested in that go to the mansion website and find your way around. Thank you very much for watching. In case you have any questions, feel free to ask them right now after this presentation or send us an email to Sebastian or me. And with that. Thanks for watching and let's see if there are any questions. Thank you very much. Bye bye.