- Integrated Systems
- About Us
- Integrated Systems
- About Us
The path to technology nirvana: Pair AI and cloud computing
Take a look at cloud computing through an AI lens. You’ll see that’s where your peers are recognizing a shift to as-a-service solutions. It’s the right move to unlock the value of your data across the enterprise with the unlimited potential of AI that can be used anywhere, anyway, and at every scale.
What do you think of when visualizing things that go well together? Delicious ice cream and apple pie. The reliable lock and key. Classic spaghetti and meatballs. Or my son’s favorite: peanut butter and jelly. And of course, I have to include the legendary Bill Hewlett and Dave Packard. The list goes on and on. Another pairing is taking hold in the technology world: artificial intelligence (AI) and cloud computing.
The role of AI in cloud computing
With advancements in technology—from the creation of smart factories that enable predictive and prescriptive maintenance to enhanced security and the expanded use of personalized medicine and image analytics, ID verification and fraud detection—the amount of data collected is astronomical.
How big is big? The total amount of data created, captured, copied, and consumed in the world is forecast to increase rapidly, reaching 59 zettabytes in 2020 and anticipated to grow to 149 zettabytes by 2024.1
Cloud computing has been instrumental in the various ways we store and access data based on priority factors such as sensitivity, volume, and velocity.
Now that you’ve collected all this data, what do you do with it? Enter artificial intelligence. AI models quickly deliver insights from large datasets and enable enterprises to make data-driven business decisions. To fully realize the value of AI—including new revenue streams and improved customer experiences—enterprises need to implement fully operational workflows. Not sure how to successfully operationalize your AI journey? Read our paper to understand the five main obstacles customers are facing in taking AI from proof of concept to production and solutions to consider.
Insightful perspectives from your peers
Recently, HPE commissioned a primary research study2 that brought cloud perceptions to the forefront and generated keen insights from your peers. The study was conducted to understand cloud perceptions and/or misconceptions, as well as identify relevant IT infrastructure trends and behaviors. Let’s take a look at some of the results through the AI lens.
It’s all about the tools
During the study, HPE asked if respondents were satisfied with their cloud strategy or if they expected to make changes. A sizable amount, at 95%, say they have plans to bring at least some of their workloads back on-prem. When digging deeper into the workload mix and singling out AI and machine learning, there is actually more of a hesitancy to move from public cloud to on-prem. There is a belief the tools and scale provided by the public cloud are superior and worth the increased costs.
The reason many implement a public cloud model is often to leverage platform services and gain access to tools. Speed, agility, along with scaling workloads also topped the list. However, what many don’t realize is the offerings in cloud models has highly evolved and perceptions from the past may not be accurate in today’s cloud landscape. There are options now for hybrid or on-prem that bring the cloud-like experience to you.
For example, you can operationalize the end-to-end ML lifecycle with a fully managed service from HPE GreenLake. It’s a turnkey, enterprise-grade solution with integrated security and a choice of data science, open-source and ISV tools. It can support any open-source tool or framework allowing enterprise data science teams to keep up with the latest innovations from the open-source community. Data scientists can rapidly deploy, test, and run ML models and collaborate with developers to analyze results. And because it’s HPE GreenLake, you only pay for the infrastructure you actually use.
On second thought, it’s all about the data
We all know that data is the lifeblood of any organization. The more data, the better. But only better by creating insights using patterns which to the human eye would be impossible. Case in point: the average clinical trial generates up to three million data points and counting.3 And with AI, you don’t need to wait to start analyzing the data, bringing the speed and agility noted earlier to practical use.
Not all data is created equal. The study looked at mission-critical workloads and how respondents defined which were most important. Not surprising, a high priority is placed on AI related workloads as the business greatly depends on insights to differentiate themselves. So what’s mission critical in the eyes of our respondents? 33% stated AI and machine learning tools, 32% identified data mining, and 31% deemed AI operations as top priorities.
Respondents were also asked about their data strategy of which majority voiced they have a mature data strategy. However, when asked: “Are you leveraging data mining and analytics using machine learning to create a data driven strategy”, the numbers fall dramatically at only 31% using AI/ML. Definitely room for advancement.
Wouldn’t it be nice if there were experts who have implemented AI successfully and could bring those learnings to you? HPE’s Artificial Intelligence Transformation Workshop can do just that. Tap into our Center of Excellence team of data scientists, solution architects, technologists, and consultants who develop and deliver advisory and professional services to accelerate your AI, data, and analytics initiatives and realize better ROI from deployments.
Not quite, it’s all about location, location, location
Business 101 classes highlight the importance of location before diving into new business ventures. The same applies with cloud model options. In the study, respondents were asked: “Where do your apps currently reside?” Through the AI lens, results showed for AI/ML 47% are in public cloud and 43% are in private. Data mining workloads are comparatively close at 45% in public and 46% in private. And lastly, AI operations at 43% in public and 44% in private. What this shows is a pretty even split of where AI related apps reside.
But a twist was found after inquiring more. Respondents were queried: “Where should your workloads reside versus where they are today?” For AI/ML, 30% of the decision makers say their public cloud workloads should be on-prem. 35% say data mining, and 33% say AI operations. Ask yourself: Why house these mission-critical apps in a location we don’t agree with?” The research revealed IT directives, offload of maintenance, lower TCO, and ease of scale among the reasons why.
As cloud models are evolving, more organizations are bringing workloads, applications and data currently in the public cloud, back on-premises and/or to hosted colocation providers. The research determined within the next two years 40% of respondents have plans to bring back 25-50% of their workloads to on-prem and another 25% are bringing more than 50% back. These are significant strategy decisions. Requirements for better security and more control over the infrastructure are leading the charge for these changes.
Correction, it’s all about the investment
We would not be doing our jobs if we didn’t investigate cloud costs. First, let’s look at investment of resources. Respondents were asked which workloads they currently are investing the most resources. The top three are:
- Structured data management at 25%
- Structured data mining/analysis at 23%
- Enterprise IT infrastructure management workloads at 21%
As honorable mentions, unstructured data mining and analysis was 18%, AI/ML, and AI operations both at 17%. Why is this important? As businesses search for ways to be more productive, efficient and get the best return, how they allocate resources plays a major part. Many are turning towards as-a-service models to help alleviate some of the resource burdens. In fact, IDC has predicted, the strongest growth in cloud revenues will come in the as a service category and by 2024 account for more than 60% of all cloud revenues worldwide, a forecast to deliver a five-year CAGR of 21%.4
An interesting point from the study examined the perceptions of the cloud’s total cost of ownership. A high percentage believes there is a lower cost in the public cloud. This belief varied by role and organization size – 57% of enterprise IT decision makers, 47% of enterprise business decision makers, 46% of SMB IT decision makers, 46% of SMB business decision makers, 61% of data scientists, and 55% of developers.
Why share all these numbers? Because in today’s IT environment, the strategy and decisions made for cloud is not on one individual. It has become a collaborative decision across the business.
And those percentages on cost perceptions of public cloud are pretty vast. But I challenge you to revisit the hybrid cloud and on-prem solution possibilities of today. HPE’s as-a-service solution, HPE GreenLake, delivers cloud economics on premises, a pay-per-use model with no capital needed up front, and no overprovisioning of infrastructure. Our portfolio has solutions for nearly all situations, data protection, containers, storage, VDI, machine learning and more.
Reality, it’s about all the above—for the full cloud experience
Honestly, the categories above—tools, data, location, investments—are all important and should be incorporated into the complete cloud experience planning. The respondents from our cloud study were asked, “What are the most important capabilities for delivering an experience from your on-premises infrastructure that is at parity with the experience delivered from public cloud providers?” The winning responses include the ability to:
- Access developer, IT Ops and analytic tools and services
- Use AI/ML services that leverage parallel processing across compute instances
- Rapidly scale up and scale down capacity as needed
Through AI, businesses are able to process data faster, extract intelligence and make data-driven decisions opening up new revenue generating opportunities. This requires thoughtful consideration about the entire data ecosystem and the right ML and AI operations platforms to manage it.
Not sure where to start? HPE Educational Services provides AI training and deep learning courses that are delivered by NVIDIA-certified instructors to ensure you get projects running smoothly.
There is no single right or wrong answer on which cloud model to implement. Your business will need to look at your individual goals and determine your right cloud mix. The key is finding a partner that understands your current stage in the AI journey—and one that can help you unlock the value of your data across the enterprise.
The unlimited potential of AI can only be realized by an enterprise if it can be used anywhere (edge-to-cloud), in any way (as-a-service or on-premises), and at every scale (device to supercomputer). To reach nirvana, look at your cloud implementation in partnership with AI.
So go ahead. Enjoy that ice cream and apple pie. Or fill up on spaghetti and meatballs. Great partnerships are just meant to be together. Like HPE and you?
Meet HPE Experts blogger Tracy Siclair
Tracy has worked for HPE for 24 years in various positions, all geared toward providing a better customer experience. She has a passion for thinking out-of-the-box and finding innovative ways to share commercial insights. Utilizing research and insights, she’s putting those into action through storytelling techniques customers can easily understand.
Hewlett Packard Enterprise
2 HPE GreenLake Challenger Research Report, HPE commissioned report, Emerald Research Group, November 2020
4 Cloud Adoption and Opportunities Will Continue to Expand Leading to a $1 Trillion Market in 2024, IDC, Oct 15, 2020