 So this is the unbedded, the SEP cluster storage solution. So who are you? I am the marketing and sales director of unbedded technology. We are a company, a startup company based in Taiwan for about 3 years. So what do you do? We are doing the SEP cluster appliance and powering by the un-base micro-server cluster which is very different from the older of the SEP product that you can see in the market. What is the SEP cluster? The SEP cluster is the software defined storage. So the foundation of the product is based on the object storage. So object storage is just like what you use in Google Drive or the Dropbox. The basement of the technology is using the object storage. What is object storage? Yeah, object storage. It's just like the distributed storage. So it's different from the black storage like you used to before. So that means you optimize in the way the cost is for hard drive and server? Is it lower cost or what does it mean if you do this solution? Okay, our solution is quite spatial and different from the others. The first thing here you see is not a powerful motherboard connected to so many hard drives. Does the ARM Cortex-M? It's the ARM V7. It's 32-bit only. But it doesn't matter because we top-up it with the software so it could make a powerful storage device. Which CPU do you have here? We are using the Marvell 385, Armada 385. So it's a dual core 1.6 GHz. Okay, dual core ARM Cortex-M? So that's the 32-bit solution right here. Why do you have so many small ones here? Because it's the micro-server architecture. So inside this 1U chassis box you have 8 micro-servers together. They have just like a cluster already. So each of the micro-servers they connect to each other. They communicate peer-to-peer 2.5 GHz per second already. 2.5 GHz per second they connect to each other inside the cluster. This is the board with the chip. This is the micro-server. So the micro-server is just like this. It's that small. It's even smaller than your credit card, smaller than your business card. And you can swap them in and out? Yeah, it's householdable. And what is this one? This one is the SSD. So we put this one for the software usage. It's just like the journal disk for our set storage. And what is this one? This one is for the remote management. So it's just like the IPMI, but we are using the auto fund manager controller. And how much bandwidth is going out? On the uplink you have 4 times 10 GB. So 40 GB? Yeah. And then you can have a lot of storage here? Yeah, we have 8 disks here. And the one thing special is each micro-server here, you connect to it's dedicated hard drive. So you can see, you can see it's just like 8 independent Linux servers inside one chassis box. And all of them, they are cluster and they are connected to each other through the onboard switch. So there's an ARM processor in each hard drive that's just driving the hard drive. And there's another one here driving the server for each of them? There's no ARM controller here. This is pure the SATA3 hard disk. Just a hard drive, yeah? Yeah, just a hard drive. And the controller is overlaid. And that's controlling, that's running a Linux for each hard drive? Yes. And then controlling all the data and making sure everything's okay and copying it? Yeah. So once you put, once you write the data into this cluster, the data will be stored distributedly over the disk you have. For example, right now you have 8 disks, right? If you stack 10 pieces of this model, then you will have 80 disks. And the data will be stored over this 80 disk, distributedly. 80? Yeah, if you have 10. Ah, 10, yeah. We could do the stack. So it's a concept for the scale out. So it's not limit. So what's the difference between this and doing, you know, like when you can, the standards that you have for copying to several hard drives at the same time? Mm-hmm. But I mean, this one is just a prototype or is ready for mass production? Oh, it's ready for mass production. We already have some customers using this for their data storage. And this concept is very nice for the massive data storage. Because currently for the send or not, they are not capable to handle such huge massive data storage. And they are not able to do the 100% scale out. But for this kind of platform, we could do the scale out very flexible. How is it different from RAID? For RAID, the data protection is doing through the RAID controller. But here, everything will be done through the software. Is it better? That's better. How much better? How is it better? If you are explaining how bad we are in terms of the power consumption, we could have to save up to 60% power consumption. And in terms of the total cost ownership, we could save at least 60% compared to the traditional ones. So who is going to buy this? Is it going to be Google and Amazon? Of course, we want to have this kind of customer. But we are starting step by step. So currently we have two different approaches. One is for the enterprise, the middle to big size enterprise. They have layer requirements for the massive data storage. And another kind of customer will be the data center. But Google or Facebook, why not? If we could reach them. And so you have a hardware solution right here. And this is the text right here. It's called Mars 200. Yeah, it's called Mars 200. And as you can see, this is exactly the idea of our architecture. We make all the microservers just like independent, just like your nerves. So they're connecting to each other by the ethernet, by the network. So we could do the cluster as big as it needs. Nice, so they connect to each other and then you have software. Yeah, we have the software. So just like I mentioned to you, this is for the, we have the dashboard. I'm sorry, I need to log in again. Okay, here we are. So for the dashboard, you can see the health conviction of this cluster. Right now we have a cluster. You have total of 14 nodes for the data storage. And the three nodes for the data monitoring is here. You have three nodes for the data monitoring. So how to scale out the cluster? You just buy the simple click. Let me show you how to do it. So CIF, it stands for? It's a name of the open source software. It's an object storage software that opens source. It's called CIF cluster storage appliance? Yes. So we provide, we made on server, on micro server, and we integrated CIF on the basement for the technology. And on top of the CIF, we develop our unified virtual storage software. This is what you see here. So eventually it's come out of CIF cluster storage appliance. So as you can see here, right now we have 14 nodes for the data storage, right? If you want to add more micro server into this service, you just simple click and tell the system, tell the cluster, which IP address you want to aggregate into this cluster, then assign the IP address and name it. And just it can be done automatically. And once the cluster is getting bigger, the data will do the auto balance. So the loading will be balanced as well. All right. Ready for mass production? It's ready for mass production. Do you have big orders already? Not very big order, but we have the data center customer already using our product to do the trial run. So they're testing it out? Yes. And if they're happy, they'll buy millions, right? Ah, I hope so. All right, cool. All right.