00:20 Deepak Mohan: Hello, I'm Deepak Mohan from IDC's Infrastructure Coverage Practice, and I lead a cloud infrastructure services research. In the next 20 minutes, we'll highlight to you some of the trends that we're seeing in terms of increasing demand for storage performance from public cloud, the use cases driving this, as well as what that means for partners in this ecosystem. We'll start with a quick look at the growth of public cloud, followed by a deep dive into those three use case segments that I mentioned, and then wrap up with how we expect this growth to evolve moving forward and the key takeaways from these presentation slides.
00:56 DM: Now, let's start by looking at the growth of public cloud. The chart that you see here is from IDC's Public Cloud Services Tracker and IDC's Cloud IT Infrastructure Tracker. The dark blue bars reflect the spend on public cloud IaaS and the light blue bars reflect the spend on enterprise IT infrastructure in terms of traditional IT infrastructure or infrastructure for private cloud fleets on premises. As you can see, in 2019, the spend on public cloud IaaS was about half of all the spend on traditional IT infrastructure. In 2020, that's expected to become about two-third, and in the next three years, we expect the spend on public cloud IaaS to, in fact, exceed the total spend on enterprise IT infrastructure for internal IT use cases.
01:44 DM: Now, the key thing to take away here is the shift in the center of gravity of enterprise IT towards public cloud. And the biggest factor driving that shift is the expanded use of public cloud. A decade ago, the primary use cases for public cloud were startups and developer use cases. This evolved to secondary storage use cases and test and dev use cases, and today public cloud is seen as an alternative to traditional infrastructure for nearly all types of enterprise IT use cases. And with this expansion of use cases has come an increase in the demands on performance from public cloud IaaS storage and compute.
02:27 DM: Now, in the next few slides, we'll discuss the three broad segments of use cases that we see driving performance demands in public cloud. The first of these are production IT applications, both business and data applications that are increasingly being deployed in public cloud IaaS. Now, one of IDC's primary vehicles for collecting empirical data on IaaS usage is IaaSView, a survey of about 1,500 enterprises worldwide collecting insights around challenges and drivers toward public cloud IaaS. Now, IaaSView reveals that one of the most frequently used use cases, or one of the most frequently deployed use cases on public cloud is, in fact, production IT applications. Now, these may be transactional data applications for customer operations, these include ERP applications for business planning, as well as various internal operations applications which are part of day-to-day business operations inside the company.
03:23 DM: Now, these aren't necessarily applications that push at the boundaries of performance availability in public cloud today, but these are applications for whom a performance degrade can directly have an impact on customer experience or on employee productivity. Enterprises moving such applications to public cloud prioritize performance and predictability and increasingly look for ways to offer an SLA and offer an experience that is consistent with what they could do on premises. We see enterprises often use SSD-backed storage with different levels of provisioned IOPS and guaranteed IOPS capabilities to deliver such use cases to their customers.
04:02 DM: As mentioned, this is reflected in IaaSView data, where we see production IT applications, both internal and external as among the most frequent applications for which public cloud is used today. This is also reflected in security concerns that are bubbling up over time. In the last four, five years, we've seen the concerns with public cloud shift from performance and skill sets to security being consistently reported as one of the top concerns that enterprises are trying to build for and design for as they build and deploy workloads in public cloud. What we believe this reflects is the increasing performance sensitivity and security sensitivity of the applications that public cloud is being considered for.
04:44 DM: Now, the second segment of use cases are SaaS software providers who want to deliver their services and their software closer to where their enterprise IT customer's workloads are. Commercially delivered software packages are expected to be an integral part of the enterprise IT environment even as they transition and build their environments in public cloud. IDC's IaaSView survey shows that over half of the customers expect to move to some type of SaaS-like consumption for these packages over time, and this is putting pressure on these software vendors to deliver their software in SaaS-like models within the public cloud environment where their customer's broader workload sit today.
05:29 DM: For these offerings, customer experience is critical, and an important part of what enables the desired predictable customer experience is the compute and storage performance that can be guaranteed on the cloud platform. As in the earlier case, software vendors delivering their services and their platforms on public cloud often prefer to use SSD-backed storage with different levels of provisioned and guaranteed IOPS and performance SLAs to ensure the customer experience and ensure a positive and responsive software experience to their enterprise IT customers.
06:07 DM: Again, this is reflected in IaaSView data where we see a variety of adjacent services that customers expect to invest in and expect to consume on the public cloud ecosystem after they migrate their initial workloads into public cloud, with databases and data analytics consistently coming up as the top areas of interest. These tend to be workloads for which storage performance has a direct correlation with the customer experience, again, underscoring the importance of storage performance for these software vendors and SaaS vendors as they deliver their services to enterprises on the public cloud.
06:45 DM: The chart on this slide is from IDC's Public Cloud Services Tracker, which tracks the spending and the forecast across the different segments of the public cloud services market -- IaaS, PaaS, SaaS and system infrastructure software as a service. Now, one of the things this shows is that the median growth in the public cloud market overall over the next five years is expected to be about 19%. During this same period, the compounded annualized growth rate for the IaaS and PaaS market, however, is 27%. Now, this higher growth in the IaaS and PaaS market are, of course, due to an increasing number of bespoke applications and enterprise-built applications being moved to public cloud, but also an increasing trend for software and SaaS providers to deliver their services adjacent to the customer's center of gravity in public cloud as we discussed earlier.
07:39 DM: The last of the three broad segments that is driving an increasing set of demands for high-performing storage in public cloud are emerging use cases, specifically data-intensive use cases like AI/ML, high-performance computing and advanced data analytics. Some of these are workloads which are moving from on-premises, but a growing percentage of these are, in fact, workloads which are cloud-first and preferring to be built on public cloud. Some of the factors driving this is the pace at which public cloud services, specifically services around high performance, in terms of GPUs, in terms of high-performance storage, in terms of adjacent capabilities that can be consumed alongside that, such as TensorFlow, NTP use and Inferentia and other capabilities of that sort have expanded and the fact that a Capex investment to support a similar capability can often be prohibitively expensive for companies that don't have that scale or that level of certainty in the duration of their projects.
08:38 DM: Now, these are use cases which are truly pushing at the boundaries of storage performance, and for the right reasons, because the storage performance and the underlying infrastructure performance has a direct correlation with the speed of machine learnings or the speed of inference possible with the algorithms and the workloads that the companies are running. So, again, going back to the data in IaaSView, we see AI/ML as one of the most and the most fastest growing areas of interest from customers who have already started using public cloud IaaS. In fact, over half of the customers that we sampled for our IaaSView survey in 2020 reported an intent to start some level of investments and initiatives around AI/ML on public cloud within the next one year.
09:24 DM: So, to summarize, these three broad segments of use cases, business and data applications that are increasingly moving to public cloud and expect a similar level of consistency and SLA capability on public cloud as they were able to deliver on premises; SaaS and enterprise software vendors who are increasingly moving to deliver software adjacent to their customer's broader workloads, who want to deliver a positive customer experience and a predictable customer experience on public cloud; and then lastly, emerging AI/ML use cases which are pushing at the boundaries of what's possible in terms of cloud storage performance today, all are increasing the demand for high-performance storage and pushing at the bar for what is expected from public cloud storage today.
10:14 DM: Now, this is also reflected in customer priorities as they evaluate public cloud providers for their workloads. Alongside factors like security and skill set transition, we see performance considerations, specifically compute and storage performance within the public cloud IaaS environment right up there among the top three factors that customers evaluate when considering a public cloud for their workload needs.
10:43 DM: And enterprises are responding in kind and . . . I beg your pardon, cloud providers are responding in kind, and we see that with the recent announcements and the recent offerings that have been introduced into the market. AWS with their io2 volumes and their E8 acquisition, Azure with ultra disks, Google with their attachable local SSDs and Oracle with their HPC instances which come with local NVMe SSDs. These are all examples of announcements in the last year specifically focused on enabling higher performance and transition of high-performance workloads into public cloud.
11:21 DM: Now, these use cases and the volume of such use cases in public cloud can only be expected to grow. Public cloud IaaS grew at 38% in 2019 and we expect the growth to stay above 30% for the next three years. One of the big factors driving this is an increasing focus on building digital capabilities both for business operations and for customer service and customer experience delivery. COVID-19 has acted as a catalyst to this by increasing the level of uncertainty in the market and increasing the propensity to accept Opex-based options, which build in flexibility and on-demand consumption as opposed to large-scale capital expenditure.
And last, but not least, are the investments from cloud providers themselves increasing the capabilities of services on public cloud and reducing the friction to migrate existing applications and build new applications on public cloud.
12:20 DM: I'll wrap up with a few . . . ways. One, we have three broad segments of use cases all of which are increasing the bar for performance expectations on public cloud. First, production business and data applications, which are either migrating to or expanding to the public cloud. Second, SaaS and software vendors who are increasingly attempting to deliver and build their software applications adjacent to where their customer's broader workloads are. And three, emerging AI/ML use cases which are pushing at the boundaries of storage performance today. All these are increasing the storage performance demands and the expectations of customers as to the storage performance that they expect or can deliver as an SLA to their downstream customers on the public cloud.
13:10 DM: Now, cloud providers are responding in kind by new offerings, and new partnerships and acquisitions to support their ability to deliver these higher performance storage needs to their customers. And what all these points to is the increasing opportunity for partners in this market to enable cloud providers to deliver higher performance storage at a lower cost and a predictable manner to their customers. And this opportunity is only expected to grow in the next five years as public cloud grows with its current CAGR of about 20% over the next five years, and customer use cases and customer demands for capabilities and public cloud continues to expand.
13:51 DM: Thank you for joining us. And if you have any questions or any topics that you'd like to discuss further, you should be able to reach me on idc.com.