Saturday, September 25, 2021

Data Storage in Multicloud: Best Practices

multicloud deployment creates numerous challenges for a company’s data storage strategy. With data-hungry applications stored on public cloud providers such as AWSGoogle Cloud and Azure, a company’s storage infrastructure and overall storage management gets exponentially more complex. Yet companies certainly must come to terms with this new complexity: multicloud is rapidly becoming the default cloud usage, and cloud itself is the foundation of enterprise IT.

In this video and podcast, we discuss:

  • What data storage challenges does a multicloud deployment create for a company, in contrast to an in-house datacenter or a single cloud provider?
  • What sort of a data storage infrastructure does a company need to best handle the demands of multicloud?
  • What are some best practices for data storage and multicloud?
  • What about the relationship between data management and storage, overall or in multicloud?
  • Future of Multicloud and data storage?

To provide insight into best practices in data storage and multicloud, I spoke with a leading expert:

Robert Lee, VP & Chief Architect, Pure Storage

James Maguire, Managing Editor, Datamation – moderator

Download the podcast:

 Video highlight: Object Storage and Cloud Native (2:10)

 Video highlight: Portability and Cloud Computing (2:24)

Full interview (24:57)

 Edited highlights from the full discussion:

Portability, Consistency and Cost in the Cloud

Lee: So I think it’s good to start with talking through why customers and why companies are balancing on-premise, cloud, hybrid cloud, and multicloud. And quite frankly, for most customers, they are starting on-prem. So the first step really is just getting to the first public cloud, which is a hybrid cloud type of deployment. And it turns out that hybrid and multicloud really presents a lot of the same challenges.

There’s lots that we can go through, but I think the three that I would call out that really rise to the top are one: How to achieve portability? How do companies avoid getting locked into one particular location or one particular cloud?

Number two is how to achieve some level of consistency. You don’t get locked into one location or cloud, but how do you get to where you can use them more interchangeably without a separate set of runbooks or a separate software stack or a separate team to have to support you in each of these environments.

And the number three that I’d call out is really cost control. How do companies avoid getting into this world of unpredictable spiraling costs.

Portability: Two Types

Lee: I would actually split portability. I think there’s actually two separate pieces of this. One is, “Hey, I’ve got an application environment machine learning data set in one cloud, do I wanna hop from cloud to cloud?” And I completely agree with you, cloud hopping, especially because of the data challenges, the cost challenges, and the network challenges, really doesn’t happen that much.

On the other hand, cloud optionality is a real thing. So customers that want to have the option to deploy part of their workload, part of their environments in cloud A, part in cloud B, and they want to be able to have that flexibility.

I’ll give you a great example of this. If you look at a lot of the SaaS providers that we serve here at Pure, these SaaS providers are providing running services to host their customers. They’re running them on some large public cloud environments. They might be built in AWS, they might have customers that refuse to store data in AWS, because of competitive reasons.

And so these SaaS providers, just as a matter of a course of business, have to be able to operate their service in multiple cloud environments in order to serve their customers. And so then now this comes back to consistency, if you’re a SaaS provider, if you want that optionality, how do you do that? How do you achieve that without having two, three different software stacks, two, three different R&D teams, two, three different support teams? And so I think, I would agree with you, cloud hopping doesn’t really happen that much, but I think everybody wants cloud optionality.

Planning

Lee: How do you balance getting the most out of the offerings that might come out in this very fast-paced, rapidly developing part of technology? How do you balance that against creating this consistency and portability?

And I think there’s a couple of things that companies can do, a couple of best practices here, right. So one is really choosing technology stacks that allow you to leverage the power of open source, right. If Google comes out with some new AI machine learning platform that really is quite powerful, it’s not gonna take very long for the open source community to embrace that to build applications, to build API’s or interfaces, if you will, on top of that.

And so as a customer, as a company, what you should be looking for is, how do I make best use of the underlying cloud environments by leveraging the power of open source, how do I make it easy to deploy open source technology?

Often, these pieces of software, microservices if you will, often delivered through cloud native stacks, through container-based technologies. And so really, again, this comes down to making your technology choices in a way that gives you maximal flexibility to leverage the best pieces. The second point I would make is, as you’re making these choices upfront and you’re building not just the technology but also the processes, it’s super, super important to focus on automating everything, right, as much as possible. If you want something to be done consistently and repeatedly, it turns out humans aren’t that great at that, computers are, right, so… [chuckle]

Cloud Native: Object Storage vs. Block Storage vs. File Storage

Lee: As you pointed out, object storage and cloud native application stacks or architectures really do go hand in hand.

Object storage plays a huge role in modern applications. You know, if we peel back the layers of the onion a little bit, this comes down to object is really uniquely well suited for these modern application environments, specifically the very data-hungry types of applications. You know, I think this comes down to really what the different types of storage were designed for.

If you think about block storage, it’s great for high-performance data access to a single or very small number of applications. You’ve got an Oracle Database, you got a VM instance, you want just very, very low latency, block is great for that.

If you think about file storage, it’s very well suited for sharing data between humans and a small number of different applications, right? I’ve got a file environment, I wanna run some applications to process data, share them across these applications, file is really well suited for that.

When we think about object storage, it’s really designed for sharing data, at huge, at massive scale across lots of compute, across lots of different types of applications. And so it’s just very well-suited for the needs of modern data-hungry types of software, but also the cloud infrastructure.

And so I think you definitely see object storage rising in importance and adoption. We definitely see this becoming a new form of primary storage. And that has significant implications. For something to be primary storage in the enterprise, there’s some table stakes, right? It’s gotta be fast.

It’s gotta be resilient. It’s gotta be secure, you gotta be able to protect it, all of the enterprise needs. And it’s gotta be consistent and available everywhere, right? So it’s gotta be available not just in the cloud, but on-premise in your data center as well.

And so we definitely see these trends certainly emerging over the last several years because of cloud native stacks, but accelerating as we look forward.

The Future: The Biggest Challenge

Lee: You know, I think certainly we’re at the phase where companies are recognizing the benefits of cloud and multicloud, but they’re also starting to see the challenges inherent.

And so I think, look, you know, the industry doesn’t move at the speed of light. It will take some time for best practices to really cement and emerge. It will take time for the requisite pieces of technology to support those best practices to really cement. And so I think you see acceleration there.

On the flip side, I think that data, data movement storage is often the biggest challenge that customers face, and so I think it will take time for customers to really navigate through that.

Similar articles

Latest Articles