Thursday, February 20, 2020

Automation and connectivity will enable the modern data center to extend to many more remote locations

Enterprise IT strategists are adapting to new demands from the industrial edge, 5G networks, and hybrid deployment models that will lead to more diverse data centers across more business settings. 

That’s the message from a broad new survey of 150 senior IT executives and data center managers on the future of the data center. IT leaders and engineers say they must transform their data centers to leverage the explosive growth of data coming from nearly every direction.

Yet, according to the Forbes-conducted survey, only a small percentage of businesses are ready for the decentralized and often small data centers that are needed to process and analyze data close to its source.

The next BriefingsDirect discussion on the latest data center strategies unpacks how more self-healing and automation will be increasingly required to manage such dispersed IT infrastructure and support increasingly hybrid deployment scenarios.

Listen to the podcast. Find it on iTunes. Read a full transcript or download a copy.
Joining us to help learn more about how modern data centers will efficiently extend to the computing edge is Martin Olsen, Vice President of Global Edge and Integrated Solutions at VertivTM. The interview is conducted by Dana Gardner, Principal Analyst at Interarbor Solutions.

Here are some excerpts:

Gardner: Martin, what’s driving this movement away from mostly centralized IT infrastructure to a much more diverse topology and architecture?
Olsen: It’s an interesting question. The way I look at it is it’s about the cloud coming to you. It certainly seems that we are moving away from centralized IT or centralized locations where we process data. It’s now more about the cloud moving beyond that model.

We are on the front steps of a profound re-architecting of the Internet. Interestingly, there’s no finish line or prescribed recipe at this point. But we need to look at processing data very, very differently.

Over the past decade or more, IT has become an integral part of our businesses. And it’s more than just back-end applications like customer relationship management (CRM), enterprise resource planning (ERP), and material requirements planning (MRP) systems that service the organization. It’s also become an integrated fabric to how we conduct our businesses.

Meeting at the edge 

Gardner: Martin, Cisco predicts there will be 28.5 billion connected devices by 2022, and KPMG says 5G networks will carry 10,000 times more traffic than current 4G networks. We’re looking at an “unknown unknown” here when it comes to what to expect from the edge.

Olsen: Yes, that’s right, and the starting point is well beyond just content distribution networks (CDNs), it’s also about home automation, so accessing your home security cameras, adjusting the temperature, and other things around home automation.

That’s now moving to business automation, where we use compute and generate data to develop, design, manufacture, deploy, and operate our offerings to customers in a much better and differentiated fashion.

We’re also trying to improve the customer experience and how we interact with consumers. So billions of devices generating an unimaginable amount of data out there, is what has become known as edge computing, which means more computing done at or near the source of data.

In the past, we pushed that data out for consuming, but now it’s much more about data meets people, it’s data interacting with people in a distributed IT environment. And then, going beyond that is 5G.
We see a paradigm shift in the way we use IT. Take the amount of tech that goes into manufacturing. It's exploding, with tens of thousands of sensors deployed in just one facility to help dramatically improve productivity and drive efficiency into the business.

We see a paradigm shift in the way we use IT. Take, for example, the amount of tech that goes into a manufacturing facility, especially high-tech manufacturing. It’s exploding, with tens of thousands of sensors deployed in just one facility to help dramatically improve productivity, differentiate, and drive efficiency into the business.

Retail operations, from a compute standpoint, now require location services to offer a personalized experience in both the pre-shop phase as well as when you go into the store, and potentially in the post-shop, or follow-up experience.

We need to deliver these services quickly, and that requires lower latency and higher levels of bandwidth. It’s increasingly about pushing out from a central standpoint to a distributed fashion. We need to be rethinking how we deploy data centers. We need to think about the future and where these data centers are going to go. Where are we going to be processing all of this data?

Where does the data go? 

Gardner: The complexity over the past 10 years about factoring cloud, hybrid cloud, private cloud, and multi-cloud is now expanding back down into the organization -- whether it’s an environment for retail, home and consumer, and undoubtedly industrial and business-to-business. How are IT leaders and engineers going to update their data centers to exploit 5G and edge computing opportunities despite this complexity?

Olsen: You have to think about it differently around your physical infrastructure. You have the data aspect of where data moves and how you process it. That’s going to sit on physical infrastructure somewhere, and it’s going to need to be managed somehow.
Learn How Self-Healing and Automation
Help Manage Dispersed IT Infrastructure
You should, therefore, think differently about redesigning and deploying the physical infrastructure. How do you operate and manage it? The concept of a data center has to transform and evolve. It’s no longer just a big building. It could be 100, 1,000, or 10,000 smaller micro data centers. These small data centers are going to be located in places we had previously never imagined you would put in IT infrastructure.

And so, the reliance on onsite technical and operational expertise has to evolve, too. You won’t necessarily have that technical support, a data center engineer walking the halls of a massive data center all day, for example. You are going to be in places like some backroom of a retail store, a manufacturing facility, or the base of a cell tower. It could be highly inaccessible.
You’ll need solutions that offer predictive operations, that have self-healing capabilities within them where they can fail in place but still operate as a function of built-in redundancy. You want to deploy solutions that have zero-touch provisioning, so you don’t have to go to every site to set it up and configure it. It needs to be done remotely and with automation built-in.

You should also consider where the applications are going to be hosted, and that’s not clear now. How much bandwidth is needed? It’s not clear. The demand is not clear at this point. As I said in the beginning, there is no finish line. There’s nothing that we can draw up and say, “This is what it’s going to be.” There is a version of it out there that’s currently focused around home automation and content distribution, and that’s just now moving to business automation, but again, not in any prescribed way yet.
You should consider where the applications are going to be hosted, and that's not clear. How much bandwidth is needed? It's not clear. There's nothing that we can draw up and say, "This is what it's going to be."

So we don’t want to adopt the “right” technologies now. And that becomes a real concern for your ability to compete over time because you can outdate yourself really, really quickly if you don’t make the right choices.

Gardner: When you face such change in your architecture and potential decentralization of micro data centers, you still need to focus on security, backup and recovery, and contingency plans for emergencies. We still need to be mission-critical, even though we are distributed. And, as you point out, many of these systems are going to be self-healing and self-configuring, which requires a different set of skills.

We have a people, process, and technology sea change coming. You at Vertiv wanted to find out what people in the field are thinking and how they are reacting to such change. Tell us about the Vertiv-Forbes survey, what you wanted to accomplish, and the top-line findings.

Survey says seek strategic change 

Olsen: We wanted to gauge the thinking and gain a sense of what the C-suite, the data center engineers, and the data center community were thinking as we face this new world of edge computing, 5G, and Internet of things (IoT). The top findings show a need for fundamental strategic change. We face a new mixture of architectures that is far more decentralized and with much more modularity, and that will mean a new way to manage and operate these data centers, too.

Based on the survey, 11 percent of C-suite executives don’t believe they are currently updated even to be ahead of current needs. They certainly don’t have the infrastructure ready for what’s needed in the future. It’s much less so with the data center engineers we polled, with only 1 percent of them believing they are ready. That means the vast majority, 99 percent, don’t believe they have the right infrastructure.

There is also broad agreement that security and bandwidth need to be updated. Concern about security is a big thing. We know from experience that security concerns have stunted remote monitoring adoption. But the sheer quantity of disparate sites required for edge computing makes it a necessity to access, assess, and potentially reconfigure and remotely fix problems through remote monitoring and access.

Vertiv is driving a high level of configurability of instruments so you can take our components and products and put them together in a multitude of different ways to provide the utmost flexibility when you deploy. We are driving modularized solutions in terms of both modular data center and modularity in terms of how it all goes together onsite. And we are adding much more intelligence into our offerings for the remote sites, as well as the connectivity to be able to access, assess, and optimize these systems remotely.

Gardner: Martin, did the survey indicate whether the IT leaders in the field are anticipating or demanding such self-configuration technologies?

Olsen: Some 24 percent of the executives reported that they expect more than 50 percent of data centers will be self-configuring or have zero-touch provisioning by 2025. And about one-third of them say that more than 50 percent of their data centers will be self-healing by then, too.

That’s not to say that they have all of the answers. That’s their prediction and their responses to what’s going to be needed to solve their needs. So, 29 percent of engineers say they don’t know what percentage of the data centers will be self-configuring and self-healing, but there is an overwhelming agreement that it is a capability they need to be thinking about. Vertiv will develop and engineer our offerings going forward based on what’s going to be put in place out there.

Gardner: So there may be more potential points of failure, but there is going to be a whole new set of technologies designed to ameliorate problems, automate, and allow the remote capability to fix things as needed. Tell us about the proper balance between automation and remote servicing. How might they work together?

Make intelligent choices before you act 

Olsen: First of all, it’s not just a physical infrastructure problem. It has everything to do with the data and workloads as well. They go hand-in-hand; it certainly requires a partnership, a team of people and organizations that come together and help.

Driving intelligence into our products and taking that data off of our systems as they operate provides actionable data. You can then offer that analysis up to non-technical people on how to rectify situations and to make changes.
Learn How Self-Healing and Automation
Help Manage Dispersed IT Infrastructure
These solutions also need to communicate with the hypervisor platforms -- whether that’s via traditional virtualization or containerization. Fundamentally, you need to be able to decide how and when to move your applications and workloads to the optimal points on the network.

We are trying to alleviate that challenge by making our offerings more intelligent and offering up actionable alarms, warnings, and recommendations to weigh choices across an overall platform. Again, it takes a partnership with the other vendors and services companies. It’s not just from a physical infrastructure standpoint.
Gardner: And when that ecosystem comes together, you can provide a constellation of data centers working in harmony to deliver services from the edge to the consumer and back to the data centers. And when you can do that around and around, like a circuit, great things can happen.

So let’s ground this, if we can, to the business reality. We are going to enable entirely new business models, with entirely new capabilities. Are there examples of how this might work across different verticals? Can you illustrate -- when you have constructed decentralized data centers properly -- the business payoffs?

Improving remote results 

Olsen: As you point out, it’s all about the business outcomes we can deliver in the field. Take healthcare. There is a shortage of healthcare expertise in rural areas. Being able to offer specialized doctors and advanced healthcare in places that you wouldn’t imagine today requires a new level of compute and network that delivers low latency all the way to the endpoints.

Imagine a truck fitted with a medical imaging suite. That’s going to have to operate somewhat autonomously. The 5G connectivity becomes essential as you process those images. They have to be graphically loaded into a central repository to be accessed by specialists around the world who read the images.

That requires two-way connectivity. A huge amount of data from these images needs to move to provide that higher level of healthcare and a better patient experience in places where we couldn’t do it before.
There will need to be aggregation points throughout the network. You will need compute to reside at multiple levels of the infrastructure. Places like the base of a cell tower could become the focal point.

So 5G plays into that, but it also means being able to process and analyze some of the data locally. There need to be aggregation points throughout the network. You will need compute to reside at multiple levels of the infrastructure. Places like the base of a cell tower could become a focal point for this.

You can imagine having four, five, six times as much compute power sitting in these places along a remote highway that is not easily accessible. So, having technical staff be able to troubleshoot those becomes vital.

There are also uses cases that will use augmented reality (AR). Think of technicians in the field being able to use AR when they dispatch a field engineer to troubleshoot a system somewhere. We can make them as effective as possible, and access expertise from around the world to help troubleshoot these sites. AR becomes a massive part of this because you can overlay what the onsite people are seeing in through 3D glasses or virtual reality glasses and help them through troubleshooting, fixing, and optimizing whatever system they might be working on.

Again, that requires compute right at the endpoint device. It requires aggregation points and connectivity all the way back to the cloud. So, it requires a complex network working together. The more advanced these use cases become -- the more remote locations we have to think through -- we are going to have to deploy infrastructure and access it as well.

Gardner: Martin, when I listen to you describe these different types of data centers with increased complexity and capabilities in the networks, it sounds expensive. But are there efficiencies you gain when you have a comprehensive design across all of the parts of the ecosystem? Are there mitigating factors that help with the total cost?

Olsen: Yes, as the net footprint of compute increases, I don’t think the cost is linear with that. We have proven that with the Vertiv technologies we have developed and already deployed. As the compute footprint increases, there is a fundamental need for driving energy efficiency into the infrastructure. That comes in the form of using more efficient ways of cooling the IT infrastructure, and we have several options around that.

It’s also from new battery technologies. You start thinking about lithium-ion batteries, which Vertiv has solutions around. Lithium-ion batteries make the solution far more resilient, more compact, and it needs much less maintenance when it sits out there.
Learn How Self-Healing and Automation
Help Manage Dispersed IT Infrastructure
So, the amount of infrastructure that’s going to go out there will certainly increase. We don’t think it’s necessarily going to be linear in terms of the cost when you pay close attention to how, as an organization, you deploy edge computing. By considering these new technologies, that’s going to help drive energy efficiency, for example.

Gardner: Were there any insights from the Forbes survey that went to the cost equation? How do the IT executives expect this to shake out?

Energy efficiency partnerships 

Olsen: We found that 71 percent of the C-suite executives said that future data centers will reduce costs. That speaks to both the fact that there will be more infrastructure out there, but that it will be more energy efficient in how it’s run.

It’s also going to reduce the cost of the overall business. Going back to the original discussion around the business outcomes, deploying infrastructure in all these different places will help drive down the overall cost of doing business.

It’s an energy efficiency play both from a very fundamental standpoint in the way you simply power and cool the equipment, and overall, as a business, in the way you deliver improved customer experience and how you deliver products and services for your customers.

Gardner: How do organizations prepare themselves to get out in front of this? As we indicated from the survey findings, not that many say they are prepared. What should they be doing now to change that?

Olsen: Yes, most organizations are unprepared for the future -- and not necessarily even in agreement on the challenges. A very small percentage of the respondents, 11 percent of executives believe that their data centers are ahead of current needs, even less so for the data center engineers. Only 44 percent of them say that their data centers are updated regularly. Only 29 percent say their data centers even meet current needs.

To prepare going forward, they should seek partnerships. Get the data centers upgraded, but also think through and understand how organizations like Vertiv have decades of experience in designing, deploying, and operating large data centers from a physical infrastructure standpoint. We use that experience and knowledge base for the data center of tomorrow. It can be a single IT rack or two going to any location.
We take all of our learning and experience and drive it into what becomes the smallest common denominator data center, which could just be a rack. These are modular solutions that are intelligent and can be optimized remotely.

We take all of that learning and experience and drive it into what becomes the smallest common denominator data center, which could just be a rack. So it’s about working with someone who has that experience, already has the data, and has the offerings of configurable, modular solutions that are intelligent and provide accessibility to access, assess, and optimize remotely. And it’s about managing the data that comes off these systems and extracts the value out of it, the way we do that with some of our offering around Vertiv LIFE Services, with very prescriptive, actionable alarms and alerts that we send from our systems.

Very few organizations can do this on their own. It’s about the ecosystem, working with companies like Vertiv, working closely with our strategic partners on the IT side, storage networks, and all the way through to the applications that make it all work in unison.

Think through how to efficiently add compute capacity across all of these new locations, what those new locations should look like, and what the requirements are from a security standpoint.

There is a resiliency aspect to it as well. In harsh environments such as high-tech manufacturing, you need to ensure the infrastructure is scalable and minimizes capital expenditure spending. The modular approach allows building for a future that may be somewhat unknown at this point. Deploying modular systems that you can easily augment and add capacity or redundancy to over time -- and that operate via robust remote management platforms -- are some of the things you want to be thinking about.

Gardner: This is one of the very few empirical edge computing research assets that I have come across, the Vertiv and Forbes collaboration survey. Where can people find out more information about it if they want more details? How is this going to be available?
Learn How Self-Healing and Automation
Help Manage Dispersed IT Infrastructure
Olsen: We want to make this available to everybody to review. In the interest of sharing the knowledge about this new frontier, the new world of edge computing, we will absolutely be making this research and study available. I want to encourage people to go visit to find more information and download the research results.
Listen to the podcast. Find it on iTunes. Read a full transcript or download a copy. Sponsor: Vertiv.

Friday, February 14, 2020

How Intility uses HPE Primera intelligent storage to move to 100 percent data uptime

The next BriefingsDirect intelligent storage innovation discussion explores how Norway-based Intility sought and found the cutting edge of intelligent storage

Stay with us as we learn how this leading managed platform services provider improved uptime -- on the road to 100 percent -- and reduced complexity for its end users.

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

To hear more about the latest in intelligent storage strategies that lead to better business outcomes, please welcome Knut Erik Raanæs, Chief Infrastructure Officer at Intility in Oslo, Norway. The interview is conducted by Dana Gardner, Principal Analyst at Interarbor Solutions.

Here are some excerpts:

Gardner: Knut, what trends and business requirements have been driving your need for Intility to be an early adopter of intelligent storage technology?æs: For us, it is important to have good storage systems that are easy to operate to lower our management costs. At the same time, it gives great uptime for our customers.

Gardner: You are dealing not only with quality of service requirements; you also have very rapid growth. How does intelligent storage help you manage such rapid growth?

Raanæs: By easily having performance trends shown so we can spot when we are running full. If that happens, we can react before we run out of capacity.

Gardner: As a managed cloud service provider, it’s important for you to have strict service level agreements (SLAs) met. Why are the requirements of cloud services particularly important when it comes to the quality of storage services?

Intelligent, worry-free storage 

Raanæs: It’s very important to have good quality of service separation because we have lots of different kinds of customers. We don’t want to have the noise-enabled problem where one customer affects another customer -- or even the virtual machine (VM) of one customer affects another VM. The applications should work independently of each other.

That’s why we have been using Hewlett Packard Enterprise (HPE) Nimble Storage. Our quality of service would be much worse at the VM disk level. It’s very good technology.

Gardner: Tell us about Intility, your size, scope, how long you have been around, and some of the major services you provide.

Raanæs: Intility was founded in 2000. We have always been focused on being a managed cloud service provider. From the start, there have been central shared services, a central platform, where we on-boarded customers and they shared email systems, and Microsoft Active Directory, along with all the application backup systems.

Over the last few years, the public cloud has made our customers more open to cloud solutions in general, and to not having servers in the local on-premises room at the office. We have now grown to more than 35,000 users, spread over 2,000 locations across 43 countries. We have 11 shared services datacenters, and we also have customers with edge location deployments due to high latency or unstable Internet connections. They need to have the data close to them.

Gardner: What is required when it comes to solving those edge storage needs?
Customers often want inexpensive solutions. We have to look at different solutions that give the best stability but don't cost too much. And we need remote management of the solution.

Raanæs: Those customers often want inexpensive solutions. So we have to look at different solutions and pick the one that gives the best stability but that also doesn’t cost too much. We also need easy remote management of the solution, without being physically present.

Gardner: At Intility, even though you’re providing infrastructure-as-a-services (IaaS), you are also providing a digital transformation benefit. You’re helping your customers mature and better manage their complexity as well as difficulty in finding skills. How does intelligent IaaS translate into digital transformation?

Raanæs: When we meet with potential customers, we focus on taking away concerns about infrastructure. They are just going to leave that part to us. The IT people can then just move up in [creating value] and focus on digitalizing the business for their customers.

Gardner: Of course, cloud-based services require overcoming challenges with security, integration, user access management, and single sign on. How are those higher-level services impacted by the need for intelligent storage?

Smart storage security

Raanæs: With intelligent storage, we can focus on having our security operations center (SOC) monitor responses the instant they see them on our platforms. We can keep a keen eye on our storage systems to make sure that nothing ever happens on the storage. That can be an early signal of something happening. Please describe the journey you have been on when it comes to storage. What systems you have been using? Why have intelligence, insights, and analysis capabilities been part of your adoption?

Raanæs: We started back in 2013 with HPE 3PAR arrays. Before that we used IBM storage. We had multiple single-Redundant Array of Inexpensive Disks (RAID) sets and had to manage hotspots ourselves, so by moving even one VM we had to try and balance it out manually.

In 2013, when we went with the first 3PAR array, we had huge benefits. That 3PAR array used less space and at the same time we didn’t have to manage or even out the hotspots. 3PAR and its active controllers were a great plus for us for many years.

But about one-and-a-half years ago, we started using HPE Nimble arrays, primarily due to the needs of VMware vCenter and quality of service requirements. Also, with the Nimble arrays, the InfoSight technology was quite nice.

Gardner: Right. And, of course, HPE is moving that InfoSight technology into more areas of their infrastructure. How important has InfoSight been for you?

Raanæs: It’s been quite useful. We had some systems that required us to use other third-party applications to give an expansive view of the performance of the environment. But those applications were quite expensive and had functionality that we really didn’t need. So at first we pulled data from the vCenter database and visualized the data. That was a huge start for us. But when InfoSight came along later it gave us even more information about the environment.

Gardner: I understand you are now also a beta customer for HPE Primera storage. Tell us about your experience with Primera. How does that move the needle forward for you?

For 100 percent uptime 

Raanæs: Yes, we have been beta testing Primera, and it has been quite interesting. It was easy to set up. I think maybe 20 minutes from getting it into the rack and just clicking through the setup. It was then operational and we could start provisioning storage to the whole system.

And with Primera, HPE is going in with 100 percent uptime guarantee. Of course, I still expect to deal with some rare incidences or outages, but it’s nice to see a company that’s willing to put their money where their mouth is, and say, “Okay, if there is any downtime or an outage happens, we are going to give you something back for it.”

Gardner: Do you expect to put HPE Primera into production soon? How would you use it first?
With Primera, HPE is going in with 100 percent uptime guarantee. It's nice to see a company that's willing to put their money where their mouth is.

Raanæs: So we are currently waiting for our next software upgrade for HPE Primera. Then we are then going to look at putting it into production. The use case is going to be general storage because we have so much more storage demand and need to try to keep it consistent, to make it easier to manage.

Gardner: And do you expect to be able to pass along these benefits of speed of deployment and 100 percent uptime to your end users? How do you think this will improve your ability to deliver SLAs and better business outcomes?

Raanæs: Yes, our end users are going to be quite happy with 100 percent uptime. No one likes downtime -- not us, not our customers. And HPE Primera’s speed of deployment means that we have more time to manage other parts of the platform and to get better service out to the customers.
Gardner: I know it’s still early and you are still in the proof of concept stage, but how about the economics? Do you expect that having such high levels of advanced intelligence across storage will translate into your ability to do more for less, and perhaps pass some of those savings on?

Raanæs: Yes, I expect that’s going to be quite beneficial for us. Because we are based in Norway, one of our largest expenses is for people. So, the more we can automate by using the systems, the better. I am really looking forward to seeing this improve and getting easier to manage systems and analyze performance within a few hours.

Gardner: On that issue of management, have you been able to use HPE Primera to the degree where you have been able to evaluate its ease of management? How beneficial is that?

Work smarter, not harder 

Raanæs: Yes, the ease of management was quite nice. With Primera you can do the service upgrade more easily. So with 3PAR, we had to schedule an upgrade with the upgrade team at HPE and had to wait a few weeks. Now we can just do the upgrade ourselves.

And hardware replacements are easier, too. We can just get a nice PDF showing you how to replace the parts. So it’s also quite nice.

I also like the part of the service processor in 3PAR that’s now just garnered with Primera; it’s in with the array. So, that’s one less thing to worry about managing.

Gardner: Knut, as we look to the future, other technologies are evolving across the infrastructure scene. When combined with something like HPE Primera, is there a whole greater than the sum of the parts? How will you will be able to use more intelligence broadly and leverage more of this opportunity for simplicity and passing that onto your end users?

Raanæs: I’m hoping that more will come in the future. We are also looking at non-volatile memory express (NVMe). That’s a caching solution and it’s ready to be built into HPE Primera, too. So that’s also quite interesting to see what the future will bring there.