 Yeah, sure hi guys. Thank you for joining me. My name is Ithai Bogner. I'm the founder and CEO of Stratoscale and for those of you attended my session yesterday it's going to be not the same but you won't learn anything new so if you have seen that I'll spare you the time and I'm going to use my Apple Watch to make progress with the slides I'm not sure it's going to work it doesn't I see okay anyway and Stratoscale is a software company we are in the emerging segment of the hyperconverged data center architecture market we are targeting enterprises and ISPs we raised 42 million dollars from Intel Cisco SanDisk Bessemer and battery we are about 60 people in the country in the in the company I'm sorry and we are based R&D based in Israel and we have a footprint in the States sales and marketing primarily and so as I mentioned we are in the hyperconverged arcade infrastructure space what we're building is a software stack that allows you to build your own cloud based on a hyperconverged infrastructure so hyperconverged infrastructure is the future of data center architectures so instead of having a split infrastructure where you have a separate storage including fabric and a separate compute infrastructure we are using server-side storage and we expose all the resources in the in a in a cluster of servers externally for your per user and we are managing we have our own software storage stack we have our own hypervisor based on KVM and we are managing the interconnect holistically the problem with hyperconvergence is that you have the storage subsystem residing on the same server as the compute as workloads running that creates interference there are both subsystems are competing for the same physical resources on every server and what we are about what we are all about is managing that interference solving contention problems and creating a very efficient and scalable infrastructure and very cost efficient because we are using a commodity hardware and local as directly attached storage inside the servers our software stack is comprised out of three building blocks the distributed storage the modified KVM hypervisor and our networking stack is managing the interconnect we are more the reason we have those three subsystems is that we have we need to have very fine grain control on what's happening inside each of the subsystems in order to manage interference and solve contention problems on top of that we have an analytics layer which is collecting runtime insights on everything that's happening inside the cluster using those insights we are able to do cluster wide load balancing and deliver a highly efficient data center to your for the workloads to be using in terms of workloads we are able to run any type of VM windows and Linux we are able to run containers we are scaling to thousands of nodes there's no scaling problems we are doing basically the only scaling limitation is the network topology because today you can build rex of servers and they are inside everywhere you have a 10 gig interconnect and it's in a mesh topology using a non-blocking switch and then we know how to daisy chain those rex into a very large data center in terms of the storage we are able to store everything so we developed our own distributed storage you do not need to manage loans you don't need to have a storage admin we're exposing a block storage device to the workloads and out of that you can carve out volumes for the workloads to be using the storage is obviously reliable it's a fully distributed and it's highly available everything is included inside our solution and in terms of the API's we are exposing to you we are exposing a cloud-like API so OpenStack API's that you are familiar with you can continue to use and build your own orchestration layer on top of our infrastructure so if you have any business logic you implemented as scripts using OpenStack API's to start instances to stop instances to connect them you can continue to use that on top of our infrastructure we expose rich analytics so again you can embed that part that logic the analytics and use that as part of your logic if you are creating your own orchestration management system we are allowing you to set SLAs per workload and policies and using that we are able to load balance the cluster and meet those SLA requirements so we are making sure that workloads do not interfere with each other and they meet your SLAs as you define them in the system in terms of performance storage IOPS networking bandwidth etc at a very high level what are the benefits of the system so for CIOs okay you get a fully distributed system it's there's no single point of failure it's a single infrastructure to serve all of your data center needs you can run any workload on top of our infrastructure it scales to a thousands of services I said for developers and architects you get an abstraction from the architecture of the infrastructure you don't need to define loans and use a storage admin to carve out storage you don't need to have a VM admin to create VMs everything is program programmable we are developer friendly and we like to think of what we are doing as a platform as a service on steroids up until the open stock API level so you get the past simplicity with the flexibility of an infrastructure as a service oh just a second it jumped too many slides in terms of something happened and I it doesn't show my slides let me see okay sorry about that so in terms of the system the system is fully distributed okay oh I talked about that I'm sorry in terms of efficiency so we allow our system allows you to get out the various efficient system out of your existing hardware okay so we know how to make create a very dense data center much more denser with current solutions we're enabling you to use commodity hardware there's no need to use a brand name servers and as a softer layer on top of the hardware we are compensating for the higher failure rates of commodity hardware so if a nose go nose go down we know how to reroute traffic we know how to use other replicates of the storage system and this allows you to save even more money by using commodity hardware I mentioned SLAs before and part of the reason we are very keen on having SLAs that helps us optimizing runtime characteristics of the workloads pair the SLAs defined by you the users so this helps us load balancing the entire clusters we do live migration of VMs we know how to throttle networking and we know how to use optimized distributed storage and reroute IOPS to the correct nodes inside the cluster to conclude we are taking an holistic approach to infrastructure we look at the resource utilization across the cluster of servers we're looking at storage compute and the networking we are the only company was built a vertically integrated solution measuring everything that is happening inside the cluster and the result of that is that we are able to load balance the cluster and deliver a very highly efficient data center for you to use we allow application architects to plan the way their applications are working because our infrastructure is predictable because we adhere to the SLAs that you define and predictability is a very important aspect of the system if you are familiar with the noise enable problem let's say in Amazon you are deploying and creating a VM you have no idea near which other VM you are going to be executing executing and how that other VM is going to impact the performance of your VM in our system because we are load balancing the cluster all the time we are preventing the noise enable problem and we are it depends on the policies we can throttle the performance of the noisy neighbor we don't allow him to use more bandwidth that is allowed so on and so forth all this requires very fine grain control of all aspects of the system storage compute and networking and and the last slide again hyperconverged infrastructure is the next generation architecture of data centers in my opinion it's going to be the dominant architectures for data centers because it delivers a very cost-efficient infrastructure high-performance infrastructure and reliability thank you very much any questions thank you