 Live from San Francisco, celebrating 10 years of high-tech coverage, it's theCUBE. Covering VMworld 2019. Brought to you by VMware and its ecosystem partners. Welcome back, this is theCUBE at VMworld 2019. I'm Stu Miniman, my co-host is John Troyer. It's our 10th year at the show. We've been going three days wall-to-wall on two sets, and really happy to welcome to the program our first-time guest, Jeff Tudor, is the Vice President and General Manager of Vision AI inside Panzora. Thanks so much for joining us. You bet, Stu, thanks, John. All right, so we've known Panzora for quite a number of years, the founder of the company, someone we've talked to a bit. I believe it's the first time we've talked about Vision AI. So maybe set the table with us of Panzora today and the value of Vision AI. Sure, absolutely. So Panzora is known predominantly for its file services, of which we can provide a global collaborative namespace across multiple different locations. So entities that are in the design, engineering, manufacturing, anything where you're working with a lot of distributed groups that need access to the same kind of working set files and big data files have been using Panzora for file services for a number of years. We're in 33 countries, 7,000 deployments, and largely in the Fortune 100. And that's kind of where we started to see that the growth of data is not only in user-generated content like PowerPoints or database backups, but it's in machine-generated data, and that's what brought us to vision.ai. Okay, so great, yeah, lay out the, that was an internally created product. How long has it been available? What's kind of the key IP in there that differentiates from others in the marketplace? So great question. Well, the thing with machine-generated data is there's a lot of it. And actually, it's growing at 60% compounding growth rate, all these great statistics. But in order to drive value of machine data, especially when you're looking at ML and AI, the larger the data set, the larger the training data set, the better the prediction models. And one of the problems with today's storage platforms for machine data is that you're taking data, you're indexing it, and then you're putting it on Flash, which is a phenomenal storage platform. But if you're looking for petabytes of Flash for just retaining a couple of months worth of data, it becomes very expensive, very fast. So a couple of years ago, we took some of the core IP that we had in creating file-to-object mapping and said, look, let's build a new cloud-native architecture to manage cloud-native digital machine-generated data and be able to transfer that, not only for the block storage and put into the object storage. So we created something called VBOS, Vision Block Object Storage, that allows us to take, index this data, and then write it to object. But still, while it's an object, have it still searchable. And that really unlocks the value of these very large data sets. So you no longer have to push this off in a tape or pushing it off in an object storage where it's no longer available. It sits in an object storage, but it's always on, it's always available. Is this a software offering, or does it sit in my bucket somewhere, or does it sit in yours? Sure, great. And then actually, are we machine-generated data? That's a pretty wide term. Are we talking log files or not? Well, certainly log files is a core starting point because that's something everybody here in the VM world has in common. As our systems of records are creating and running virtual machines, it's generating the digital data about who accessed, what, when, when, where, when, and how, for IP addresses, security information, dashboards, et cetera. So we've created this as a service because in a multi-cloud world, you need kind of one platform where you can ingest these data feeds and these log feeds, and then store that and search it. People have been generating and deploying on-site log files for some time, but we've seen a large interest among our customer base in a hosted service that can securely store and make their logs accessible. All right, Jeff, bring us inside a customer. What's some of the kind of typical use cases, outcomes that customers, if you have any example that might illustrate it, I'd love that. Sure, so we'll take a customer that is in the publishing business. And as you know, in the publishing business, we were going from paper into digital, right? So it's this digital transformation. And as their industry changed, they became now a web hoster. So the sites when there are papers that used to advertise and they're classifieds and buy print ads, they're now managing their digital experience. Well, as they're doing that, they came into a situation where some of their sites were having unpredictable performance loads. And they're not a really, they're just sophisticated enough to have one IT person managing 50 different to about 50 different servers, virtual machines running these hosting these sites. So they needed help. Is there a platform that can come help me create dashboards so I can visualize this log data that came into us? So our partner, one of our key partners here is Phoenix Snap at the show where in Intel's booth demonstrating our Optane Accelerated Technology. So we went into this particular customer, onboarded them in five minutes, created the dashboards for them and now their logs are coming in at a number of gigabytes per day and that can visualize and find out any points of their operations that are creating problems and slow access time for their customers. You know, I love the storage and data aspect of it, right? The searchable object store sounds very neat. I bet there's some very cool computer science there. Storage and data geeks love that. It's also got AI in the name and we talked about ML and AI. So where does that come into the picture? Absolutely, so great question. The AI and ML aspect of this is because as you get primarily the large data set sizes, then you can start putting machine generated algorithms on top of it, right? So creating large data stores and then the first machine generated analytics that we've run on top of it are things such as kind of storage prediction costs. It's actually one where we've saved one of our customers in the financial services tens of thousands of dollars a month because we can look at their bucket, their bucket sizes and the access time so they're S3 buckets and say look, you know, you're actually not accessing it. You can drop it down into infrequent access and you're not going to get a higher bill. So we can run these analytics for them and provide that data to them. All right, Jeff, we're here at VMworld. VMworld's talking a lot about multi-cloud and microservices, cloud native, VMworld cloud pieces. Help us understand the intersection between what you're doing and how that ties into VMwear and their customers. Absolutely, well, in a multi-cloud world, VMwear is obviously one very important component of it. But there's also components that are non vSphere based, right, and so we have to be cognizant of this and need a platform that can support any data feed from any data source. So that's certainly one of them. But number two is you mentioned it, the microservice. Traditional log platforms or machine data platforms, such as Elasticsearch or Splunk or things like that, is where you go and you create your architecture and your infrastructure and you manage that infrastructure as you're putting that data into it. So it puts operational burden on the customer to go manage all this. In our view of the world, it needs to be completely serverless. You need to be able to consume machine data, log data as a microservice. In a complete serverless methodology. So you send your data into this URL, it goes into your buckets, encrypted, it's dropped into your object service where it's searchable. Yeah, it's funny. I've been looking at the serverless space for a couple of years now. Functions, really interesting stuff. Kelsey Hightower actually put out, he said, isn't most of networking serverless by definition, maybe just clarify that for us. Yeah, so serverless, it's just like the cloud. It's just somebody else's data center. I actually have the teacher serverless that updates that there is no cloud. It's a computer, it's just a microservice that you pay a little bit for when you need it. When you need it, right? But really it gets into, if I want to spin up Elasticsearch, let's talk about that, because that is one of the key workloads that's running in our platform. When you talk about Elasticsearch service, if you want to spin that up, you need to go literally spin up virtual machines, assign block storage to those virtual machines, and hope that you assign enough storage for your data ingestion. Otherwise your performance is going to go down, your data is going to become blocked, and you're going to need to assign storage. So you're still managing stuff, even though it's in the cloud. In our world, we're kind of trying to turn machine generated data and democratize it into simple as a Gmail account, right? I create, I request a microservice endpoint, then you write to that endpoint. Now of course we're managing servers, and we're managing clusters and virtual machines, and all that funness, but it's transparent to you, and most importantly, you're not hit with any cost for the infrastructure. You're only charged for what you're consuming, and that means it's a complete consumable base model from that standpoint, which saves customers a lot of money from otherwise having to buy and host a lot of infrastructure. So Jeff, you have a big presence here at the show, a nice booth, I hope it's going to be a good week. I'm curious what you thought the energy was, kind of, I think you all had an announcement. I'm not, but I could talk a little bit about that and how that works with the ecosystem and the audience here. Yeah, we've actually, we have two announcements, and let's take the first one, the file services. Because from our file service platform, we're announcing vCent certification, which is coming in the fall. We've gone through that process so that anywhere you're running a VMware on any of the cloud providers on top of sand, vSAN, you now have a file services platform on top of that that can expand beyond just your NVMe and also leverage that object storage for this kind of infinite, you know, filer, if you will, for that. But the other announcement we have is the log analytic service. All right, yeah, tell us a little bit about customer meetings you're having. What are the things that are kind of bringing customers to you? Is there a certain thing that, you know, when you hear it, you're like, ah, this is a perfect Panzer or a customer? Well, yeah, certainly. I would think that any, you know, data and storage is just a universal problem and people can't get enough of it and ultimately they want to get out of the business of managing storage in a large case. So in that particular instance, being able to offer them a software defined file system platform for our traditional filer environment is something that's going to, it's just an evergreen force, right? It's going to continue to grow, you know, the performance of file and flash at the price of object. That's a pretty clear value proposition. In the machine-generated data analytics space, which vision.ai, you know, it is, how do I make sense of my data? I need to take all of these data streams and actually put some intelligence on it and create alerts, you know, visualize this data. So our big proposition here is five minutes to visualize your data and that resonates. I can, you know, walk these customers that are traditionally having to go build their own long-service environments and I'm saying here, let me onboard you, we can actually start sending their data up and having visualizations and alerts in five minutes and that's revolutionary to them, right? It's just the simplicity of it is key and I think that's making IT simple to consume and democratizing is something we're focused on doing. Jeff, the last thing I have is, tell us a little bit about what we should expect going forward. Obviously the AI and ML stuff is continuing to grow. What should customers be looking for from Panzora in the near future? No matter how sophisticated a customer in an enterprise is, they don't have enough smart people, right? And data scientists are very expensive and they're very scarce. So what we are doing and focused on doing and we will be doing more of is we've built a marketplace, a marketplace where data applications, data analytics applications can be created by the community, can be published into and be consumed by an enterprise. So that they have their account, they add in this application, they can immediately start utilizing and experiencing and unlocking the power of their data. Jeff Tudor, really appreciate the updates on Panzora. Congrats on the progress of Vision AI and hope to catch up with you in the future. Thanks so much, yeah, look forward to next year. For John Troyer, I'm Stu Miniman, getting towards the end of our coverage from VMworld 2019, but as always, thanks for watching theCUBE.