Wednesday, February 10, 2021

How consistent storage services across all tiers and platforms attains data simplicity, compatibility, and lower cost


T
his
BriefingsDirect Data Strategies Insights discussion series, Part 2, explores the latest technologies and products delivering common data services across today’s hybrid cloud, distributed data centers, and burgeoning edge landscapes.

New advances in storage technologies, standards, and methods have changed the game when it comes to overcoming the obstacles businesses too often face when seeking pervasive analytics across their systems and services.

Stay with us now as we examine how IBM Storage is leveraging containers and the latest storage advances to deliver inclusive, comprehensive, and actionable storage. 

Listen to the podcast. Find it on iTunes. Read a full transcript or download a copy.

To learn more about the future of storage strategies that accelerate digital transformation, please welcome Denis Kennelly, General Manager, IBM Storage. The interview is conducted by Dana Gardner, Principal Analyst at Interarbor Solutions.

Here are some excerpts:

Gardner: In our earlier discussion we learned about the business needs and IBM’s large-scale vision for global, consistent data. Let’s now delve beneath the covers into what enables this new era of data-driven business transformation.

In our last discussion, we also talked about containers -- how they had been typically relegated to application development. What should businesses know about the value of containers more broadly within the storage arena as well as across other elements of IT?

Containers for ease, efficiency

Kennelly: Sometimes we talk about containers as being unique to application development, but I think the real business value of containers is in the operational simplicity and cost savings.

Kennelly

When you build applications on containers, they are container-aware. When you look at Kubernetes and the controls you have there as an operations IT person, you can scale up and scale down your applications seamlessly.

As we think about that and about storage, we have to include storage under that umbrella. Traditionally, storage was independently doing of a lot of the work. Now we are in a much more integrated environment where you have cloud-like behaviors. And you want to deliver those cloud-like behaviors end-to-end -- be it for the applications, for the data, for the storage, and even for the network -- right across the board. That way you can have a much more seamless, easier, and operationally efficient way of running your environment.

Containers are much more than just an application development tool; they are a key enabler to operational improvement across the board.

Gardner: Because hybrid cloud and multi-cloud environments are essential for digital business transformation, what does this container value bring to bridging the hybrid gap? How do containers lead to a consistent and actionable environment, without integrations and complexity thwarting wider use of assets around the globe?

Kennelly: Let’s talk about what a hybrid cloud is. To me, a hybrid cloud is the ability to run workloads on a public cloud and on a private cloud traditional data center. And even right out to edge locations in your enterprise where there are no IT people whatsoever.

Being able to do that consistently across that environment -- that’s what containers bring. They allow a layer of abstraction above the target environment, be it a bare-metal server, a virtual machine (VM), or a cloud service – and you can do that seamlessly across those environments.

That’s what a hybrid cloud platform is and what enables that are containers and being able to have a seamless runtime across this entire environment.

Today, as an enterprise, we still have assets sitting on a data center. Yet typical horizontal business processes, such as HR or sales, want to move to a SaaS model while still retaining core differentiating business processes.

And that’s core to digital transformation, because when we start to think about where we are today as an enterprise, we still have assets sitting on the data center. Typically, what you see out there are horizontal business processes, such as human resources or sales, and you might want to move those more to a software as a service (SaaS) capability while still retaining your core, differentiating business processes.

For compliance or regulatory reasons, you may need to keep those assets in the data center. Maybe you can move some pieces. But at the same time, you want to have the level of efficiency you gain from cloud-like economics. You want to be able to respond to business needs, to scale up and scale down the environment, and not design the environment for a worst-case scenario.

That’s why a hybrid cloud platform is so critical. And underneath that, why containers are a key enabler. Then, if you think about the data in storage, you want to seamlessly integrate that into a hybrid environment as well.

Gardner: Of course, the hybrid cloud environment extends these days more broadly with the connected edge included. For many organizations the edge increasingly allows real-time analytics capabilities by taking advantage of having compute in so many more environments and closer to so many more devices.

What is it about the IBM hybrid storage vision that allows for more data to reside at the edge without having to move it into a cloud, analyze it there, and move it back? How are containers enabling more data to stay local and still be part of a coordinated whole greater than the sum of the parts?

Data and analytics at the edge

Kennelly: As an industry, we go from being centralized to decentralized -- what I call a pendulum movement every number of years. If you think back, we were in the mainframe, where everything was very centralized. Then we went to distributed systems and decentralized everything.

With cloud we began to recentralize everything again. And now we are moving our clouds back out to the edge for a lot of reasons, largely because of egress and ingress challenges and to seek efficiency in moving more and more of that data.

When I think about edge, I am not necessarily thinking about Internet of things (IoT) devices or sensors, but in a lot of cases this is about branch and remote locations. That’s where a core part of the enterprise operates, but not necessarily with an IT team there. And that part of the enterprise is generating data from what’s happening in that facility, be it a manufacturing plant, a distribution center, or many others.

As you generate that data, you also want to generate the analytics that are key to understanding how the business is reacting and responding. Do you want to move all that data to a central cloud to run analytics, and then take the result back out to that distribution center? You can do that, but it’s highly inefficient -- and very costly.

What our clients are asking for is to keep the data out at these locations and to run the analytics locally. But, of course, with all of the analytics you still want to share some of that data with a central cloud.

So, what’s really important is that you can share across this entire environment, be it from a central data center or a central cloud out to an edge location and provide what we call seamless access across this environment.

With our technology, with things like IBM Spectrum Scale, you gain that seamless access. We abstract the data access as if you are accessing the data locally -- or it could be back in the cloud. But in terms of the applications, it really doesn’t care. That seamless access is core to what we are doing.

Gardner:  The IBM Storage portfolio is broad and venerable. It includes flash, disk, and tape, which continues to have many viable use cases. So, let’s talk about the products and how they extend the consistency and commonality that we have talked about and how that portfolio then buttresses the larger hybrid storage vision.

Storage supports all environments

Kennelly: One of the key design points of our portfolio, particularly our flash line, is being able to run in all environments. We have one software code base across our entire portfolio. That code runs on our disk subsystems and disk controllers, but it can also run on your platform of choice. So, we absolutely support all platforms across the board. So that’s one design principle.

Secondly, we embrace containers very heavily. And being able to run on containers and provide data services across those containers provides that seamless access that I talked about. That’s a second major design principle.

Yet as we look at our storage portfolio, we also want to make sure we optimize the storage and optimize the spend by the customer by tiered storage and being able to move data across those different tiers of storage.

As we look at our storage portfolio, we also want to make sure we optimize the storage and optimize the spend by the customer by tiered storage and being able to move data across those different tiers of storage.

You mentioned tape storage. And so, for example, at times you may want to move from fast, online, always-on, and high-end storage to a lower tier of less expensive storage such as tape, maybe for data retention reasons. You’ll then need an air gap solution and you’ll want to move to cold storage, as we call it, i.e. on tape. We support that capability and we can manage your data across that environment.

There are three core design principles to our IBM Storage portfolio. Number one is we can run seamlessly across these environments. Number two, we provide seamless access to the data across those environments. And number three, we support optimization of the storage for the use case needed, such being able to tier the storage to your economic and workload needs.

Gardner: Of course, what people are also interested in these days is the FlashSystem performance. Tell us about some of the latest and greatest when it comes to FlashSystem. You have the new 5200, the high-end 9200, and those also complement some of your other products like ESS 3200.

Flash provides best performance

Kennelly: Yes, we continue to expand the portfolio. With the FlashSystems, and some of our recent launches, some things don’t change. We’re still able to run across these different environments.

But in terms of price-performance, especially with the work we have done around our flash technology, we have optimized our storage subsystems to use standard flash technologies. In terms of price for throughput, when we look at this against our competitors, we offer twice the performance for roughly half the price. And this has been proven as we look at our competitors’ technology.

That’s due to leveraging our innovations around what we call the FlashCore Module, wherein we are able to use standard flash in those disk drives and enable compression on the fly. That’s driving the roadmap in terms of throughput and performance at a very, very competitive price point.

Gardner: Many of our readers and listeners, Denis, are focused on their digital business transformation. They might not be familiar with some of these underlying technological advances, particularly end-to-end Non-Volatile Memory Express (NVMe). So why are these systems doing things that just weren’t possible before?

Kennelly: A lot of it comes down to where the technology is today and the price points that we can get from flash from our vendors. And that’s why we are optimizing our flash roadmap and our flash drives within these systems. It’s really pushing the envelope in terms of performance and throughput across our flash platforms.

Gardner: The desired end-product for many organizations is better and pervasive analytics. And one of the great things about artificial intelligence (AI) and machine learning (ML) is it’s not only an output -- it’s a feature of the process of enhancing storage and IT.

How are IT systems and storage using AI inside these devices and across these solutions? What is AI bringing to enable better storage performance at a lower price point?

Kennelly: We continue to optimize what we can do in our flash technology, as I said. But when you embark on an AI project, something like 70 to 80 percent of the spend is around discovery, gaining access to the data, and finding out where the data assets are. And we have capabilities like IBM Spectrum Discover that help catalog and understand where the data is and how to access that data. It’s a critical piece of our portfolio on that journey to AI.

We also have integrations with AI services like Cloudera out of the box so that we can seamlessly integrate with those platforms and help those platforms differentiate using our Spectrum Scale technology.

But in terms of AI, we have some really key enablers to help accelerate AI projects through discovery and integration with some of the big AI platforms.

Gardner: And these new storage platforms are knocking off some impressive numbers around high availability and low latency. We are also seeing a great deal of consolidation around storage arrays and managing storage as a single pool.

On the economics of the IBM FlashSystem approach, these performance attributes are also being enhanced by reducing operational costs and moving from CapEx to OpEx purchasing.

Storage-as-a-service delivers

Kennelly: Yes, there is no question we are moving toward an OpEx model. When I talked about cloud economics and cloud-like flexibility behavior at a technology level, that’s only one side of the equation.

On the business side, IT is demanding cloud consumption models, OpEx-type models, and pay-as-you-go. It’s not just a pure financial equation, it's also how you consume the technology. And storage is no different. This is why we are doing a lot of innovation around storage-as-a-service. But what does that really mean?

It means you ask for a service. “I need a certain type of storage with this type of availability, this type of performance, and this type of throughput.” Then we as a storage vendor take care of all the details behind that. We get the actual devices on the floor that meet those requirements and manage that.

As those assets depreciate over a number of years, we replace and update those assets in a seamless manner to the client.

We already have the technology to support all environments. Now we want to make sure we have a seamless consumption model and the business processes of delivering storage-as-a-service and being able to replace and upgrade that storage over time -- all seamless to the client.

As the storage sits in the data center, maybe the customer says, “I want to move some of that data to a cloud instance.” We also offer a seamless capability to move the data over to the cloud and run that service on the cloud.

We already have all the technology to do that and the platform support for all of those environments. What we are working on now is making sure we have a seamless consumption model and the business processes of delivering that storage-as-a-service, and how to replace and upgrade that storage over time -- while making it all seamless to the client.

I see storage moving quickly to this new storage consumption model, a pure OpEx model. That’s where we as an industry will go over the next few years.

Gardner: Another big element of reducing your total cost of ownership over time is in how well systems can be managed. When you have a common pool approach, a comprehensive portfolio approach, you also gain visibility, a single pane of glass when it comes to managing these systems.

Intelligent insights via storage

Kennelly: That’s an area we continue to invest in heavily. Our IBM Storage Insights platform provides tremendous insights in how the storage subsystems are running operationally. It also provides insights within the storage in terms of where you have space constraints or where you may need to expand.

But that’s not just a manual dashboard that we present to an operator. We are also infusing AI quite heavily into that platform and using AIOps to integrate with Storage Insights to run storage operations at much lower costs and with more automation.

And we can do that in a consistent manner right across the environments, whether it’s a flash storage array, mainframe attached, or a tape device. It’s all seamless across the environment. You can see those tiers and storage as one platform and so are able to respond quickly to events and understand events as they are happening.

Gardner: As we close out, Denis, for many organizations hybrid cloud means that they don’t always know what’s coming and lack control over predicting their IT requirements. Deciding in advance how things get deployed isn’t always an option.

How do the IBM FlashSystems, and your recent announcements in February 2021, provide a path to a crawl-walk-run adoption approach? How do people begin this journey regardless of the type of organization and the size of the organization?

Kennelly: We are introducing an update to our FlashSystem 5200 platform, which is our entry point platform. Now, that consistent software platform runs our storage software, IBM Spectrum Virtualize. It’s the same software as in our high-end arrays at the very top of our pyramid of capabilities.

As part of that announcement, we are also supporting other public cloud vendors. So you can run the software on our arrays, or you can move it out to run on a public cloud. You have tremendous flexibility and choice due to the consistent software platform.

And, as I said, it’s our entry point so the price is very, very competitive. This is a part of the market where we see tremendous growth. You can experience the best of the IBM Storage platform at a low-cost entry point, but also get the tremendous flexibility. You can scale up that environment within your data center and right out to your choice of how to use the same capabilities across the hybrid cloud.

There has been tremendous innovation by the IBM team to make sure that our software supports this myriad of platforms, but also at a price point that is the sweet spot of what customers are asking for now.

Gardner: It strikes me that we are on the vanguard of some major new advances in storage, but they are not just relegated to the largest enterprises. Even the smallest enterprises can take advantage and exploit these great technologies and storage benefits.

Kennelly: Absolutely. When we look at the storage market, the fastest growing part is at that lower price point -- where it’s below $50K to $100K unit costs. That’s where we see tremendous growth in the market and we are serving it very well and very efficiently with our platforms. And, of course, as people want to scale and grow, they can do that in a consistent and predictable manner.

Listen to the podcast. Find it on iTunes. Read a full transcript or download a copy. Sponsor: IBM Storage.

You may also be interested in: