 Okay. Hello everybody. Sorry for that. So I'm Julian Kazeiro, co-founder of OpenIO. I also work on OpenStack integration and all the ecosystem around OpenIO. So OpenIO is an open source object storage solution. It's a pure software and the company was created in June 2015 but the solution, software solution was created ten years ago to handle large emails platform for Telco in France. So we bear in France, we have also an office in San Francisco in Tokyo also. Okay. So at OpenIO we aim to be really easy to install and manage storage and to use it. So what can you do in five minutes? Of course you can drink an espresso, maybe stretch your body and deploy an OpenIO cluster. So easy. Don't have to be really an expert in that. So everything we do at OpenIO is to bring massive storage, massive scale storage to your data center and to be easy. So what about the compute and storage demo that we all face when we have a lot of compute and storage resources? Usually you end up with silos of independent, component storage resources and what you end up with is wasting resources. So you over-provise things to keep up with SLAs and QoS and then the costs are completely above the roof and the complexity becomes nightmare. So what OpenIO can do about that? The idea of OpenIO is that it's pure software, you install it on commodity hardware, so any regular Linux server you can find, you install our software on top of that and then you can create a really hyperscalable storage pool. You choose your hardware, which is really great because if you have older hardware, you can break it into the pool and mix hardware together. We know how to handle such case. It's really easy to start your cluster this way. So what OpenIO can do about it and how we do it? So to store, OpenIO really handles really large scalable storage. You can have literally billions of objects in a single cluster and thousands of petabytes of data. It's really easy to retain such numbers. There's a very simple shell-nothing architecture. Just bring independent servers into the pool and you can build a limitless storage pool out of regular standard servers. All the metadata is distributed by nature and it's really easy to have millions of containers as in Amazon S3 buckets, very similar, and you have billions of objects storing on top of that. Of course, storage for the protection is what really matters with OpenIO and with this object storage solution. What you can choose is between data replication and of course, the more efficient errors you're coding techniques that you will probably want to use. Everything is supported out of the box with OpenIO and you can also build more complex topologies where you have multiple data centers across different zones or maybe build a stretch cluster. That's something really easy to do with OpenIO. Since the object storage layer is usually accessible with HTTP-like protocols, we of course support Amazon S3 and OpenStack Swift integration. It's really easy to have huge read and write total throughput of bandwidth with these kind of architectures. Every node can serve the data and can write the data. We also provide file sharing protocols on top of OpenIO. You can use it with NFS for example and it stores on the same grid and processing. This is what OpenIO does a bit differently than the others. What we want to do is if you have a large pool of servers like that, you usually end up wasting RAM or CPO because the storage software doesn't really use it. You can run applications on top of the cluster and you are closer to where the data is and where your software will interact with it. It's really a simple architecture and of course to reduce the cost because you don't have to buy any more servers on the side to handle your workload. For example, what applications we run on top of OpenIO is indexing for full text search on the objects. What you can do also is video transcoding. Imagine that you have a workflow and you put software video streaming and you want to transcode it, which is really easily done with OpenIO, just bring transcoder into the grid and you have a large scalable video transcoding grid capability. But really how we do it, how does it work? The main idea is directory. You don't want to track objects because there's billions of them. So what we do is we track containers. So containers is just a huge index and you can have millions of them on the same cluster and containers track objects. So it's a really convenient way to handle such a large scale and it really works. So the structure is flat. You can't have a container on top of another. It's just everything is flat like Amazon S3 buckets, also with containers for that matter. This is really what makes it different than Swift, for example, or Ceph. They usually end up with consistent ashing techniques or some variations of it. The idea of OpenIO is that you should not rebalance data when you grow the cluster. It's really not good stuff to do in production because you end up wasting time moving data around and you can't really use your new hardware on the go. So what we do at OpenIO is that when you add new nodes, they are automatically discovered and they are immediately available. And there's no static data placement, no ring or anything. So it's really easier to manage this way and to scale. For the storage pool, how we choose where to put the data since it's not consistent ashing based on static data placement. What we do is that we have a process called the Conscience. The Conscience knows every node of the grid and it knows how it behaves and its performance. So we call it, every time we have the opportunity to do so, metrics. We compute a score with that and we distribute it to all the clients and all the nodes of the grid. That can make us do dynamic load balancing. So when you put the data, you really want to be sure that the node you're contacting is up, that it has good performance, anything like that. So how do we compare to other kind of data center platforms regarding storage? Of course, we are hardware agnostic and most open source software are. We don't rely on any appliance. You can really choose your hardware. It's open source. So really, the TCO is pretty low and we have a low cost per gigabyte per month kind of model. So it's really easy to start with OpenIO. Actually, you can start right now. I don't have a license to get. What we provide at OpenIO is licensing. It's not licensing, but it's support. Sorry. Compared to hyperconverged platforms, OpenIO really scales at web scale like billions of objects and petabytes of data. So what about the use case? What can it do with OpenIO? OpenIO started with email and has an email storage software solution. So it can really handle email workloads very well. We have integrations with various e-mail servers such as SIRIS, IMAP, or Zimbra. You can use OpenIO with a brand road over use case such as media content processing, so for videos or for data archiving. OpenIO made the OpenStack integration. We provide Swift integration. So it relies on the Swift proxy. So we use the Swift proxy from OpenStack and we integrate our native Python API into the Swift proxy. So it's not based on the driver on the object server side. You don't need any object servers or any account servers or any container servers from Swift. It's really just the Swift proxy. Which is great because you can use the Keystone authentication, the same as OpenStack does, and it's compatible with OpenStack Swift middleware. So for example, the static web middleware can be used to handle this kind of static website storage. And of course, it's open source. It's on GitHub. You can check it out. And recently, OpenStack became a corporate member of the OpenStack foundation. So I'm really, really, really happy to be part of the OpenStack family. We are really young in the family, but we are actively working on, of course, the object storage side, but we are also focusing on the final system, shared integration of Manila. So Manila is something we are really looking into. And what about the latest things we did with OpenIO? So we released our new software. It's aligned with OpenStack Summit in Austin. So 1604. 1604 brings NFS. We bring also HTTP gateway for video streaming. You can stream videos right from the cluster. We also have support for ARM, and you can run a cluster on Raspberry Pi, which is a really small computer. So you can have 10 nodes of OpenIO running object storage like that. We also provide packages for all the main distributions out there, Linux distribution. And recently, OpenIO became a member of the Kinetic OpenStorage project. We're a member of the Linux foundation, and we're really excited to be part of this Kinetic group. We think that it can really change the way we handle storage today and how you build massive storage platforms. So thank you very much for your attention. Come visit us at A16 booth. It's right over there.