Community Podcast: Nutanix Object Storage Service

  • 1 August 2018
  • 0 replies
Community Podcast: Nutanix Object Storage Service
Userlevel 7
Badge +35
Welcome back. This week on the podcast, I chat with Rohit Goyal, Product Marketing Manager for Nutanix and Sharon Santana, Technical Marketing Engineer for Nutanix about Nutanix object storage services. I want to give you an update on the offering, some insights into what we have planned. We talk about trends, data sprawl, what is Nutanix, object storage, what are the early use cases, and we have some fun at the end. Let's join the conversation.


Thank you Angelo.

I thought we'd start by talking about some of the industry trends and what we're seeing as it relates to data sprawl and usage.

Sure. So a couple things to note Angelo, is what you want to do is you want to look at some industry trends and see what they're talking about. So according to IBC, by 2020, total data volume will hit a mind boggling 40 zettabytes with almost 63 percent of that data sitting outside structure databases. And if you think about that growth, it's really about the flexibility, scalability and [inaudible 00:01:47] of management, ways to preserve and protect that data. So you need a solution or you need an architecture that will enable you to handle these data volumes.

Yeah, that's a lot of data. And I recall when working in enterprise IT, object storage was never something on our radar. But actually, now that I think about it, maybe it was our file servers that were acting as an object storage service. I remember times when our file servers would run out of space ans the team would spend cycles removing MP3s, executables, etc.

I think what's happened in the past is, a lot of folks are really used to some of the SAND technologies out there, so your traditional block, storage erase, and they're also used to these filers. So they look at these NAS solutions. But what really is changing the game is how much data is growing and then, where are you storing that data? So, a couple of things to keep in mind when you're thinking about storing a lot of this unstructured data, meaning your video files, your media files, MP3s, things like that, you really don't want to store it on those expensive disc erase, right? Think about where it's optimally placed and where you can put it so that you have it when you need it, but it doesn't have to be on the low latency discs that your traditional database workloads are accustomed to.

Yeah, I couldn't agree more. I feel like as we continue to evolve in the technologies that we are experiencing these days and creating and producing, the expectations are changing. Now we don't have patience to wait around. A second feels like an eternity, right? So as we continue to consume all of these gadgets and all of this data keeps getting produced, you're gonna see the need for something like objects more and more out there, because it allows you to have the flexibility to manipulate this data in the petabyte scale with ease.

Why don't we take a moment here to define object storage and also explain why it's different than block or file storage?
Sure. So traditionally, primary data is stored as blocks and files within your data center. Storage is organized in the way it's accessed. If it's accessed at the raw block level, then that's typically called, "Block storage." Block storage can either be directly attached or shared over the network. If you're thinking about the latter, those are typically called a SAND, or Storage Attached Network Device. SAND devices use iSCSI and fiber channel protocols. And if you think about file, those typically use NFS or SNB, formally known as SIFTs. Files are typically stored on NAS devices or Network Attached Storage devices.

These types of storage are naturally found close to compute resources, but the data as it continually grows, they're just not ready to be migrated to cold storage and doesn't need to be stored close to that compute. There should be options for storing that data efficiently, securely and cost effectively.

So object storage is designed for really, that unstructured data that's highly scalable and resilient for the world of file computing. It's really not in a fixed format or consistent, basically consistent varying sizes. It's storage that's accessible over the network via simple S3 compatible HDP rest API calls. It's based on the defacto standard that Amazon web services started back in 2006 on the S3 APIs. Most of the object storage providers out there today have that. And it's a really well documented interface. So a lot of folks have adopted it and so have we.

Earlier this year at .NEXT New Orleans, you gave a presentation on Nutanix object storage. Can you speak to some of the early use cases and some of the functionality we'll see in object storage?

Yeah, sure. So I hope a lot of you were able to attend .NEXT, first of all. It was a great show and we put on a US show every year around the May timeframe, and then we do another show in the MIA region, so for all of you folks who are listening out there in the MIA region, we do put on an annual show towards the end of the year. And we really put together, across the board, we really take inputs, what folks wanna learn, and we keep the show technical.

So in this session, we definitely jumped into talking about why we built it in the first place and what's going on. So it goes back to that traditional architecture that we were talking about. And instead of replacing silos, what we wanna do is we want you to think about, like in architecture, and how you can enable services as you need. So for example, we introduced files services a couple years back, and it's a service that you can enable through software. So once you've kind of adopted the Nutanix software architecture, know that through simple software upgrades, you have that capability of enabling something like file services, block services, or object services.

And it's not a discussion around replacing like for like. So when you're looking at your data center, really think about it holistically in how you manage it, right? So it's not that you want to go and troubleshoot all of these different components and figure out all the complexity between them, it's really about having that right foundation, that right architecture, and being able to set up these services across the board.

And then what we talked about in the session was what you can do with object storage, right? And as data was growing, what we needed to do as a company is figure out a way to solve that problem for our customers. And to do that, we have to enable service that allows them to store this data in a manner that's cost effective and that scales, right? And then on top of that, being able to leverage data services that we have that are built in. So the erasure coding, the deduplication, the compression algorithms that you're accustomed to having. We need to have those built in. And those are all built on web scale technology that Nutanix has. So that's a big differentiator for us.

And then on top of that, being able to scale that out. So, we've got customers running over two thousand notes. So you can definitely start small and you can scale it out. And you can handle lots and lots of different types of data from a few terabytes or even kilobytes all the way to some petabyte range in that object storage. So think about billions of objects being able to be stored in something that's a massively scalable system that's a single main space and allows for things to scale.

And then on top of that, we have that security built in. So we have software based encryption, we have hardware based encryption, and these are really great for those healthcare protocols, financial protocols, federal. So if you're in those vertical scenarios, think about how we have that capability of keeping your data safe. And maybe Sharon can hit on a couple of the use cases that we have.

So for the first release of Nutanix object storage service, we're targeting or prioritizing, so to speak, three main use cases. And these will be DevOps, long-term data retention and backups. Of course, there is more plan on the road map, but for now, let me give you a few examples of what these mean.

So object storage service will be loaded with features to support these use cases that I just mentioned. In the case of DevOps , which by the way, refers to the integration of development and operation teams within an organic station and it seeks to produce more robust products and applications by unifying software development and software operations. It is important because this kind of practice is in cultures, they incorporated a lot of automation. And it is important for them to have agility and flexibility.

So targeting DevOps, for us, means that we can offer them, like Rohit said earlier, single global name space with a time to first idle 10 milliseconds or less. And the fact that we support Amazon's S3 API which is a well-known and understood protocol today means that engineers or these teams can access this data using a familiar interface with potentially little or no change to the code that they are using today if they were to move to object storage tanks.

Another example would be long-term data retention as I said earlier. And this would have features that would include, for example, write once, read many, or aka, "Worm policies." And these worm policies help users of the object storage on the tanks to meet regulatory compliance. When such policies are applied to entities such as pockets, objects or tags, it prevents the data from being changed or tampered with or deleted, even. So this is really important for organizations in the healthcare, financial, or government sectors.

And also, there's gonna be support for new versions. So it's not like you couldn't ever alter an object, you could keep uploading as you need, but you could keep those later versions as you keep uploading more objects. And what this prevents is for people or users who accidentally override or get rid of the data that can't be tampered with or changed. So those are a few examples as well as data backups I mentioned, object storage service tanks will support integration to third party backup services.

Yeah. I think that's great. Just a couple things to add there, really, around some of the points that you brought out, one is that machines are generating data these days, right? And with these machines generating the data, it's not only about the files and the media files and MP3s and things like that that humans are generating, it's really around this machine generate data and I think Sharon does a great job of saying, "Hey, where are you gonna put that?" and you wanna store it, you're not ready to toss this data out the window. So how do you do it? And you really need a solution, again, that's gonna scale with you when you need it.

In terms of architecture, I know during your presentation, you had a architecture diagram,, and I thought maybe you could describe it for folks as best we can through audio. What are the pieces involved in object storage?

I'll take that one. So at a high level, Nutanix object storage is integrated right into the solution stack via at least what we call, "Object volume manager," so at least one of those. And an instance of object storage can be confirmed by one or more object volume managers. And object volume manager, you can think of them as a logical volume manager that has multiple components. And together, they provide the object storage API and handle the copies for the backend storage for the object.

We also have a scale out modular design in which we have all of these components separately. So each component focuses on a single core function so that it can do it better and can do it well. Any of these components can be scaled out independently so that they can match the workload demands of the particular deployment. So that makes it scalable and flexible to fit the needs of what the object storage instance is being utilized for. And despite that layered approach, we can still also leverage unique performance optimizations to streamline the data path. So, what that means is that the object volume managers running on top of the CVNs can store hints about the physical location of the meta data at the object volume manager layer. All this means is that it facilitates faster reads since the hints can be stored. And it keeps latency for time to first byte low.

I know you did a demo of this at .NEXT, in terms of availability, what are we looking at?

Yeah. So what we're doing is we're pushing ahead. And we did show a couple demos, you're right at .NEXT and in previous .NEXTs, and we were showing integrations with companies like HYCU, backing up directly to our object storage service using the S3 API natively, meaning you didn't have to change any of your application or workloads, they just natively wave the right in terms of backing up, and then they were able to read and restore right from our object storage.

So we did it with HYCU, Commvault and we're gonna keep pushing ahead. And we're targeting later on this year, Angelo. So really excited about getting this out the door and to our customers and prospects.

I know you also shared a road map during that session at .NEXT. Any items you'd like to highlight on the podcast?

A couple things you wanna keep in mind that we were building into the object storage service that we're gonna GA later this year is around the warm pockets that Sharon highlighted. So being able to do the right ones, read many capabilities where nobody can delete that data, not even Nutanix ... So we know that data, once it's set for the warm policies, that it'll be set and you'll be able to meet that compliance, whether it's healthcare, financial, or government regulations.

And then we're thinking about object versioning, so being able to create copies of objects and protect your data from accidentally overriding or deleting it.

And then we're thinking about object tagging, quickly identify objects by tagging them. And you can think about your projects that aren't meeting compliance, again, and being able to tag that data.

And then what we also wanna do is have that petabyte support. So when you start small and you scale, know that Nutanix will be able to solve that for you. So we know a lot of object storage providers out there, they require you to start at a half a petabyte and then they make sense and then they scale up from there. Well, we truly do start small. So you can start as small as you'd like and then you can scale with us, right? So remember that fact that I was talking about earlier where you have that capability of doing over two thousand notes.

And then we're also thinking about multi part upload, being able to reduce those upload times by breaking that data into these chunks and being able to quickly upload those documents or images or videos into that truly single global main space.
And much more, I mean we're gonna constantly iterate on this and produce as much as we can to make sure that the product is fully big for you folks.

Another cool thing to keep in mind is that unlike other public cloud providers, there will be no cost for data increase when using Nutanix object storage. So that could be huge savings, 'cause we're talking about a lot of data. So terabytes to petabytes of data. Something to keep in mind.

Great points. Let me just quickly highlight three areas where you can learn more and get involved. One is the online community at Great place for community blogs, a lot of forums where you can ask questions and continue your conversation around different topics. Another is attend a Nutanix user group in your local area, great way to get connected and familiar with folks in the community. And I also encourage you to introduce yourself to the local leader, find ways of getting more involved. And if there's a topic idea of interest, share that with the leader. We'll do our best to get someone out there to talk on it and or produce content that the leader can share at your next meeting.

And finally, we have a new area on the Community for Product Ideas. You can submit a product idea and the community begins to vote on it. As the idea gets up voted, depending on how many people in the community are interested in that particular idea, it begins to move through a work flow where ultimately, it gets to the hands of Nutanix Engineering and they'll evaluate and decide whether it's something that enough people are interested in, and it just might make it into a future product. So you never know. So I encourage you to check out those resources.

Anything additional you'd like to share, Sharon?

We also have a new school channel on YouTube where we post what we call Tech Topic videos. There will be some object storage videos uploaded there in the near future going over the object storage service feature overall as well as a demo that you can look forward to. There's also great resources there to find out more information about any of the other Nutanix solution features that we mentioned like compression or deduplication and of course, as always, the Nutanix bible for more information on how those things work. So keep that in mind if you want to learn more about those.

Awesome. Yeah, so definitely follows us on Twitter. We do post things on Twitter that make sure that we're giving you the latest and greatest information. And we're actively working on our website, so we're always gonna be keeping that up to date. So go to and then look for the object storage offering there and you'll definitely get all of the information that we talked about today and much more.

Great. And as we close out the podcast, when I have special guests on, I like to surprise them with two rapid fire questions. They don't know what they're going to be asked. So here we go. Sharon, IOS or Android?

Android all the way.

Favorite app on your Android phone?

I would say Instagram. Sorry.

Awesome. Rohit, IOS or Android?

IOS. I have one and I can't really get off of it.

Favorite app on your iPhone?

I would say WhatsApp. It helps me keep in touch with family abroad.

Well there you have it. Thanks for joining us. We'll see you on the next podcast. Thanks for joining me today, be sure to check out the Nutanix online community at for resources, blogs, and continuing the conversation in our forums.

A reminder that .NEXT London will be happening in November. If you're a blogger, ping me and I can provide a blogger pass. If you have a show idea or would like to be a guest on the show, I can be reached at community at From the team here at Nutanix, have a great week.

0 replies

Be the first to reply!