 Hi everybody, my name is David Svarmak, or Chairman, depends on what country you're coming from. I want to show you what we did, what was our journey, we took those things in the last three or four years. And you may be wondering why I froze it down like, I mean, let me show you who we are first. We've been doing those tests on the market since 1991, so we've also been doing it for five years. We have a new internal learning group of 20 million in Europe, so we have a great big company. And we actually have two divisions. One division is Infrastructure Services, so we operate roughly 18 data centers in central Europe, not all data centers, but these data centers. And we run different services in those data centers, like hardware. We develop different solutions in the house as well. And the second part of that is the value of the distribution, the value of the reseller of different brands. We are a pretty large disco partner. We do various networks, which is in 29 countries in central Eastern Europe. So we have two divisions. However, there is a huge synergy between those two divisions. We sell hardware or we use it as well, so we can help our customers. We show them how to do that. We have hands-on experience. We like to save the funds, so we can use that in our infrastructure division as well. Cloud services was the logical next step, because we were selling hardware to our customers, and they were housing or collating the hardware and the data centers. So we decided to build some virtualized solutions. We tried different ones. So we decided to use some open solution, not to be another VR shop. And we selected OpenStack roughly three or four years ago as our full solution. We started to develop some things around it, like our small ecosystem, and first production customers we launched two years ago on I-SAMOS. So what is the actual infrastructure we're using for OpenStack? We have 18 data centers. A good example of our cloud infrastructure is Prague. In Prague we have three data centers. In those three data centers, we built one OpenStack cluster. It's actually more OpenStack cluster, so we sold private clouds, posted private clouds. But they are always located in three data centers. There is a high availability between those data centers, so we can actually withstand the downtime of one data center without losing any data from the system start. Yes, in the other data center. So this is the cloud infrastructure in Prague. We opened new POPs during this year. Let's say ultimately we are starting US, Hong Kong and Dubai. So what's the current OpenStack status? I'm not talking about Mitaka or any other version. OpenStack is clearly an industry standard. So there is no other open source or open cloud social market with so much drive from an industry standard. It's a really good framework. It's not product. You really need to build a huge ecosystem around that. But the framework is good. What's bad? It's still far away from the employee use. There are so many new projects and it's growing horizontally. However, it needs to be done in the most important projects like Nova. So it's a great chance for us to be the company who will actually make it running for 4 customers. What customers want? They are already using VMware in-house virtualization using KMAs and target space Azure public clouds. But they want private clouds locally. They really need someone who will work with them to make the infrastructure better and more general. They want transparent pricing. You get pretty transparent pricing amazingly, but the pricing will go sky high. So private clouds can save you a lot of money when you have large infrastructure. We focus on larger customers from 5 to 10 Ems. We focus on customers from 500 to 1000 Ems more. Where we actually see the next step is platform as a service. Customers, at the moment, they don't want that. But we are pushing our current IS customers to start to use PAS. Because in PAS, you get much better TCO. You can save a lot of money on operations. You can be much more agile, flexible applications without additional costs. And of course, who the customers want the most they want help with the cloud transition. So giving them just the infrastructure is something else. We actually told at the beginning that we will give the infrastructure to our customers we will spend some time with them and they will start to use it. After these two years we know it's not like that. We spend much more time with customers than we thought. So what the customers can use for private clouds they can use DIY with your solved approach. They can use public clouds to give them a private cloud solution which are very flexible at this moment. A lot of suppliers are using heavy machinery. So really huge complex systems of heavy easter tasks. And there are a lot of service provided in the market. VMware shops which are moving to OpenStack and they see that they can save money the customers can save money but they have no background and they can't just click next next next how they did that with VMware it doesn't work with OpenStack. And there are other non-standard based service providers and there are many of them who have been using it for years so they are not based on OpenStack they don't have APIs. So what are the gaps? Where are the gaps we identified and how we are leveraging them? Actually we want to bring something to customers with large infrastructure which means they are paying a lot of money for that and we want to show them how they can save money they are paying every month but how they can save money on long term, long run What we need to do for customers is actually bring prepared product which is ready deploy it for them and provide services on top of that. What are the needed skills to run a production-grade OpenStack or any cloud service? You need to know all this every single part of this and you need to mention this If you don't have it you will fail or you will fail at some moment We learned in the last few years that you can't contract third parties to handle these things because at the end we are selling infrastructure services to our customers there are tight SLAs and we experienced with our partners and we paid SLAs the partners didn't pay anything or they paid this small part we got issues with our customers our cooperation was ruined or not ruined but it hurt What we found out is that the only way we can really do this is to have enter and know how to company in-house, to really have engineers to be able to have all these things in-house It doesn't mean that we need to do this when it comes to development or long-term development it's great to have partners and work with them I would like to cover comfortable topics These are probably the most important topics when it comes to OpenStack I'm going to cover those four topics because we have only 30 minutes or 25 minutes so why not to go fast Customers are looking for trust so what they need is trustworthy product let's say OpenStack is trustworthy they are looking for trustworthy service provider and they are looking for trustworthy infrastructure of course they trust in their application so it's like a chain of trust customers want trust we want trust as well we need trust in our infrastructure and this is something I'm going to cover today so what we did to provision to deploy infrastructure we decided to do fully automated environment build fully automated environment so we actually have continuous integration continuous delivery system for the provisioning as well for the development of the deployment process we are not fixing anything by hand so if anything breaks it doesn't matter if it's hardware it's like a software issue we just trash the hypervisor we trash everything we really want from scratch which gives us more stability there are no hand fixed things we know that every single node is running this application in this state for every single cloud deploy we have single cloud definition tree structures in the era and using this definition you can build the cloud deploy provisioning, operate it expand it, downgrade it upgrade anything you need to do you have this single source of trust in that in the definition file OpenStack release I'll start with this small box OpenStack release is a new version every six months every version goes end of life after 12 months usually sometimes sooner than 12 months 11 months so you need to do upgrade every new version if you don't do that and you want to skip one version you will actually start to use end of life versions of OpenStack and there could be issues with the upgrade path it's really good to upgrade to every new version what we did we built wall solution in Wcontainers so every single part which can be run in Wcontainer is running in Wcontainers that brings us really good abilities to upgrade, downgrade anything to do snapshots of configurations snapshots of databases we can do security between services so we can set files in containers we can scale containers we run one, two, three, five of them none of those containers has a configuration inside so containers are actually using this single point of source of information which was in the last slide so we can start new container and do anything we can upgrade all back to old infrastructure and it actually brings us much better security than running every Wcontainer or in EMS this is something which is pretty there is no other company on the market when it definitely comes to OpenStack who can offer this and quite the time working in trusted computing so because we want to trust in our infrastructure we want our customers to be sure that we trust in our infrastructure so so we are using trusted computing DCG standard as in standard DBM, Intel TXT to be able to trust in the infrastructure when this hardware comes we can safely discover them provision them and we can operate them as well at every point of time we know that this is the hardware we ordered at the beginning the configuration is correct, BIOS is ok operating system is the kernel which is running there is what we want there nobody have tampered with any of these parts and we can decommission the hardware because using this technology we encrypt hard drives as well it is actually really interesting because even our administrators they can't access the keys which are using for the hard drives so if we decommission hard drives we can just take them on service and we don't trade them we don't need to take care of anything what does it mean they are not going to pay fine because of not working hard drives and trust computing is actually pretty difficult to understand technology is almost ready there are so many different things you can do with this but nobody is using it nobody is really using it we really started from the hardware to decommission and we know that we can trust the infrastructure trusted logging, another thing which is completely missing in OpenStack unfortunately logging is missing in OpenStack a little bit there is still a lot of things which needs to be done in OpenStack to really provide the fucking data to service providers or to OpenStack users what we did with logging was decoding and we did sealing so every single log we know that this log has been made at this time with this hardware and we signed it with some RSA key and nobody can tamper with logs so when it comes to compliance you know that all your logs you have in your database are actually the original logs there are a lot of other things when it comes to security in OpenStack operating system limited privilege escalation GR security to be added to factory authentication which is something which probably with the next OpenStack series you will have it there but I mean everybody has to factory out authentication it's standard on this data encryption so you need to encrypt data between data centers at last not between service and records you need to do this between data centers in terms of protection and detection systems which can actually protect you against zero-day attacks because I mean you have pretty huge APIs in OpenStack and when you have more services you have bigger API base so you need to protect yourself against security attacks and for us accessing the infrastructure of a customer it means we need to add some security on our side so we're using hardware tokens different things but it's called authentication monitoring we actually came what I didn't tell you in the beginning is that we spent two years working on this we are finally coming out of the stealth mode we were really working on the stack and we are now trying to sell it to customers and to show it to the public we actually came to the point that we're going to open source some of our products because we think that customers like to write for customers to have some of these things so we're going to open source some things when it comes to monitoring and performance you need customers to see what is how the cluster is performing not just what their VMs are working but how the storage is performing, what the latency is what's the switching time between the VMs how the hypervisors are doing so this is what we built in thousands and thousands of metrics on any subsystem there what you see here is a monitoring for SAP storage this is one part we're going to open source with this you see all information about SAP how the VMs are working subsystems hypervisors, OSDs what's happening there I mean every single thing and this is again something which doesn't exist storage for infrastructure you need block storage volumes you don't need anything else you can build any other kind of storage on top of that but this is the basic thing you need you need software defined solutions you don't want to buy games in attack because with software defined solutions you buy servers, you buy which connects to those servers and drives and you can install this kind of software as a software defined storage after one or two years there's a new software solution which is much more flexible faster it's much better and you can just reinstall and you have it and this is actually under heavy development there are really new products in the market and you can use software defined solutions we tried many of them LB and block storage which are not scalable you can really can't roll with storage somehow iSCSI finding redundant software defined iSCSI is impossible task roster and many of them too complex, we all support different things so everybody is in SAF SAF is the standard when it comes to open stack it's super slow, CPU hungry there's a really good benefit of SAF that you always have your data if you configure it properly you always have your data of course you can't access them fast and when it comes to issues it's really difficult to repair SAF and we've been using SAF for two years so we know exactly what this is about plus with SAF it's really hard to get 24x7 fast support you can get support but when the storage goes down you need response time like this ok, i said two new software defined roles ok, another contribution to open source community from our company DRVD, you probably know DRVD everybody who ever built highly available open source solution definitely came across DRVD DRVD is a standard for that it's 15 years in production and we are working with Lingbit Austrian company with the creator of DRVD to bring it as a new super fast scalable reliable storage to the open stack and everybody who knows SAF will try this and we actually did some experiments 40x faster than SAF using the same CPU resources same SSDs and high devices 40x faster it's not 5x faster it's really huge improvement and when it comes to storage ok, you have your data but you need to backup your data and when you are a sales provider of course you want to make sure that your customer data are so safe with DRVD it's much easier than with SAF we can do agentless backup we were already doing this with SAF as well but it was like huge amount of code but with DRVD we can do this much faster we can actually backup every single volume agentless in the background we can save those volumes same storage save it offline or offsite or inside and we can actually bring APIs to our customers so they can recover those saved backups volumes to new volumes they can access it from using their common open stack APIs so for sales provider it's must you really need it ok, so this where main topics when it comes to open stack and IS we can at this moment we're still selling IS to our customers that's the largest testing base we spent last year working on the platform as a service we never sold this product to any new customer we always take our IS customers we convince them ok guys I want to save some of your time it's better to use platform as a service and we actually over the last year 15 months now we work with our customers to build a new platform as a service solution where the customers can actually develop deliver tests and operate their applications using Git as a code repository and a similar port like this so for the customer and it's built on top of affection new source, VCOS something like very new technologies and what you can do with this you can really save a lot of money with IS you need to still take care of your VMs you need to take care of scaling of different things so in the past you don't need to do that you go to the dashboard create a new application and for inside that dashboard you attach different modules to the application which means like database, page, engines radius and you connect this project to the Git repository and from that moment you control everything using Git so you can configure those modules in Git you can put specific configurations, scaling requirements, things like that you version those things so you can go to a new version you can roll back to the old version you can use this dashboard to build your own environment so if you have 5 developers every single developer can use his own environment, deploy new versions in that environment and when he is finished with the development he can click send it to his manager and his manager, the project manager can actually print that version and push it to the production which is just one click and in production environment all the scaling is there so this is this is something pretty unique again why? because there are some products in the market there are some service providers in the market offering these things a company which can do it end to end so from the service software development a company can build new modules for you up to the the layer when it comes to data centers and we are running the production environment so and this is something which is still pretty new customers you know they are okay with this but you know it's getting we are getting better but let's say IS is still there but in 2, 3, 4 years the market will be definitely moving towards this direction and for us we need to be there because when you stay with IS in 3 years IS can be gone won't be gone but you know there will be so many new competitors and not much development when it comes to new functional modules there is platform as a service I just put the screenshot here we have a booth there we will come to us if you want to see something that's pretty much it I want to say one more thing we built all this around OpenStack so we are not modifying OpenStack code using vanilla versions or you know we are actually using kind of a little and we are building everything around that which is great because when we were from from we just we just upgraded the OpenStack we changed a couple of things around that but we don't use this proprietary versions we are not modifying the code inside which means OpenStack is pretty framework as I told you at the beginning if you build around this you can upgrade your framework and it will help you to build your around services ok my time is up do you have any questions we still have one minute maybe ok so thank you very much for listening