Resultados 1 a 2 de 2
  1. #1
    WHT-BR Top Member
    Data de Ingresso
    Dec 2010

    [EN] Why There’s Still Room for Cloud Solution Startups

    Joe Liebkind
    October 2, 2017

    A recent Cowen study revealed that companies will be spending 16 percent more on public cloud services in 2017 year-over-year. Organizations are also increasing their use of cloud resources in their work. 18 percent of workloads are now supported by public cloud and 77 percent of current cloud users are likely to add a Software-as-a-Service (SaaS) workload in the next two years. Gartner also expects the public cloud market to grow 18 percent this year.

    Amazon Web Services (AWS) and Microsoft Azure continue to lead the market and are poised to continue their dominance. The Cowen study revealed that AWS and Azure lead all facets of user experience including user interface, billing, support, and software and data integrations. AWS remains ahead but Azure is starting to chip away at AWS’ lead as Azure displays its savvy in the enterprise market.

    This dominance of big-name tech companies should come as no surprise. AWS had the advantage of being the first to offer computing resources through a subscription model. Microsoft and IBM both have their longstanding presence in enterprise computing and Google will always have its following among developers and startups. So, in this game of giants, one may wonder if there’s room for other players to participate.

    By the numbers, it appears that there should be plenty of room for competition. The total cloud services market is still worth $246 billion this year and is expected to be more than $300 billion come 2020. Marketing wisdom has it that ventures should find a need and fill it. There are specific customer needs that aren’t necessarily covered by these major cloud providers much like how niche and boutique businesses could address focused yet lucrative segments. In the 2017 version of its Magic Quadrant for Cloud Infrastructure-as-a-Service (IaaS), Gartner included a number of niche players such as hosting provider Rackspace and Pan-European service provider Interoute in their listing indicating that there’s room for cloud services that focus on specific verticals.

    For instance, independent software vendors (ISVs) that seek to streamline their infrastructure requirements can do well exploring these premium cloud services. Unlike major providers, these niche players offer turnkey solutions features that cater to specific uses such as development and testing, user management, and automated provisioning.

    ForgeRock, a digital identity and access management service, tapped cloud provider CloudShare to leverage virtual lab management capabilities. ForgeRock uses cloud-based virtual labs to be able to train users in the full capabilities of their solutions. Using such a service provides them with key advantages. Aside from virtual training, CloudShare’s virtual labs are used in delivering sales demos and proofs-of-concept and as development and test labs.

    Many of these small to medium-sized ISVs often place a premium on being able to do more with less. For instance, environment provisioning is not necessarily a complicated process thanks to improvements in virtualization technology but it can take time when done manually. If there is need for dozens of similar virtual environments and machines to be created, the process can take hours or days to finish when attempted by a single resource. Through turnkey solutions, a user may only take a few minutes to a few hours to accomplish a similar task.

    Also in demand by today’s more agile ventures are turnkey cloud platforms for application development. Cloud platform service Heroku, for example, has gained a following among smaller development teams since the service allows them to create various environments that support different development frameworks easily. This allows them to experiment and even use different programming languages. By taking away the burden of managing the environments and the nitty-gritty of software configuration, developers can focus on the more critical and creative aspects of development.

    It’s not just startups or smaller ventures that can benefit from premium features. ISVs that offer solutions requiring network complexity (such as cybersecurity) would also find such functionalities helpful. For instance, demonstrating enterprise applications has its challenges. Traditionally, sales teams would often have to carry or ship their own hardware to perform on-premises demos. This method also has its shortcomings since it is difficult to showcase multiple use cases through limited hardware. Using cloud resources, more immersive demos are possible since teams could simply create multiple environments and set them up as part of complex network topologies – something currently not readily available in most major cloud providers.

    Often, the savings from person-hours alone can justify the price of these services may command. Fortunately, the as-a-service model adopted by cloud service providers has eased the need for upfront costs and companies only pay for the resources they actually use.

    Major cloud service providers tend to focus on the infrastructure level and creating services that cover a broad spectrum of use cases. The nuanced requirements of companies inevitably would create gaps in their service lineup allowing premium services to focus on these specific needs. As more companies realize that shifting to a cloud infrastructure may not be as straightforward as tapping a major provider, they will ultimately be drawn towards these niche players that could offer them convenience and solutions to their problems.

  2. #2
    WHT-BR Top Member
    Data de Ingresso
    Dec 2010

    Why Adoption of Public Cloud Won’t Exceed 17% of IT Demand by 2022

    Mark Thiele
    October 3, 2017

    There are over 300 hyperscale data centers in the world, with more being added every month. Each hyperscale facility is estimated to have between 50,000 and 80,000 physical servers. The numbers alone should provide significant shock value, but when considered in combination with compound growth, Jevons paradox and new technology trends, it should really start to blow your mind.

    We read and hear a broad spectrum of data points and opinions on what the future might be for colocation growth or how much of the infrastructure market will be in public cloud. What we don’t see or hear is what the growth looks like and what the reality of market dynamics, current IT trends and simple capacity management issues might mean to how we grow.

    The Cloud Service Providers (CSPs) can’t build fast enough

    The CSPs are some of the premier data center designers and builders in the world. They have teams larger than many mid-sized companies just working on the creation of new data center capacity. In fact, many of the larger wholesale colocation companies and even some retail ones are used by the CSPs as outsourced just-in-time capacity.

    The law of numbers is both awesome and scary at the same time. Consider that our major cloud providers have roughly 13.65 million servers globally (AWS, Azure, Alibaba, Baidu, NTT, etc.).* The expectation of continued server growth is that there is a close correlation between revenue growth and the size of the infrastructure. I’ve decided for this article to use 25 percent CAGR for all cloud providers through 2022, even though Amazon (AWS) reported growth of 43% in Q1 2017.

    Although I don’t have an exact figure, my industry experience combined with conversations with luminaries like Derek Collison, CEO of Apcera (@derekcollison), Tim Crawford, CIO advisor at AVOA (@tcrawford) and other senior execs from the CSPs who can’t be named, I believe that current market penetration for CSPs is roughly 10 percent of the Total Addressable Market (TAM). Assuming 10 percent market penetration of public cloud is correct, it would mean there are approximately 136.5 million servers in use globally. Keep in mind that the CSPs generally use their server resources much more efficiently than enterprises do, so it’s possible there are more like 160 million+ servers in the world.

    *Lydia Leong of Gartner estimates between 50k & 80k servers in each hyperscale facility. With 300 facilities at an average use of 70% capacity and an average max capacity of 65k servers, that makes 13,650,000 servers (65k x 300) x .70

    What are some of the drivers for this growth in CSPs & colocation?

    A not insignificant portion of growth in the CSPs has been net new. Many of these new workloads belong to new startups, but a big portion belong to new opportunities associated with the many trends happening in the technology market right now. So, we have a combination effect of Jevons Paradox on the increased use of IT due to more efficient and cost effective acquisition and the reality that many CIOs see the value of having ITaaS.

    Trends that are likely to have significant impact on IT demand over the next 5 years

    Artificial Intelligence (AI/ML)

    • You can’t read three technology articles without running into an article involving AI. As cost of infrastructure goes down and the capability of solutions improve, the growth of AI will only accelerate. It’s safe to say that few functions in the modern world are safe from AI.
    • I see AI driving many new business models that would otherwise not have been feasible (growth of overall IT demand), but I also expect that though AI created efficiencies many legacy systems will be replaced/retired. The net effect is likely to be more IT overall, but not necessarily a massive increase.

    Virtual Reality (VR)

    • Like AI there are an endless set of opportunities for VR. Customer interactions, travel, scenario planning, construction, manufacturing, gaming, etc. VR, like AI, will likely displace many current systems, but there will also be a very large number of net new applications that there is no existing (however inefficient) substitute for.

    Internet of Things (IoT) in all it various forms

    • IoT will drive millions of new use cases from making processes more efficient (deliveries, manufacturing, etc.) to building smart systems that allow for better interaction with facilities and the weather. The key though - getting closer to the customer – this is where IoT will drive real value for adopters. A very large percentage of these new IoT-enabled IT systems will be net new.

    Drones & Autonomous Vehicles

    • While drones and autonomous vehicles will have systems that use AI, VR, IoT, big data and cloud, they are also new systems that will create new business models requiring net new IT. In many cases though where there is a displacement the displacement will be in the labor force (drivers, farm workers, photographers, interns for scientists, etc.).


    • Blockchain will drive so much cost out of many businesses that it will inspire investment and growth in other areas of the business.
    • There will be millions of new services offered that will benefit from blockchain in providing trust and efficiency when it is most sorely needed.


    • I saved cloud for last because it is definitely not least. It’s ironic that cloud growth will be positively impacted by each of the above trends while at the same time it’s also enabling those trends through the reduced friction of acquisition.

    What does all of this mean for IT?

    The entire IT market will likely nearly double by 2022 and we can expect to see a threefold (3X) growth of cloud and the associated data center footprint globally by 2022 and yet CSPs will still only makeup approximately 15-17 percent of total IT infrastructure in use. This is of course great news for the cloud suppliers who continue to win in this market, but it means several other things.

    • For CSPs to get to ~ 15 percent penetration of the infrastructure market by 2022, they will have to grow threefold in four to five years. Imagine if the CSPs were going to get to 50 percent of IT (as some suggest) infrastructure TAM by 2022. They would have to grow 7-8X in five years. That kind of growth isn’t feasible in a market where the data center building experts of Microsoft, NTT, Alibaba, and AWS are already struggling to keep up with demand.
    • Enterprises will continue their trend to free up capital, improve performance, increase agility and drive up resiliency through the process of reducing their reliance on internal data centers. However, since they can’t put everything on the public cloud, even if they want to, it means the colocation market stands to continue to benefit for many years to come.
    • The colocation market in the US is estimated to be 20 percent of the current data center footprint. If they double in five years they will still only be 20-25% percent of the market as the TAM will be double the current size. I expect to see the colocation market continue to grow at 15-20%+ CAGR for the foreseeable future.

    Expected Distribution of IT infrastructure in 2022

    Percentage of opportunity is more important that the number of servers

    The use of servers as a unit of capacity continues to get more complex to define. Are we talking about CPUs, sockets or physical boxes? What about virtual machines or containers? If we were to measure based on containers, Google might already have over 100 million servers on its own. If we’re talking virtualization, then we would need to easily multiply my server estimate by five times. What about thinking of servers as a socket? A socket could have one CPU or 16, or a socket could be an ARM, FPGA or a GPU or multiples of the same. For the sake of this little exercise, I’m just thinking about growth in demand coupled with historical experience.

    The Opportunity

    As the costs continue to go down and the capabilities continue to improve, humans will continue to find new ways to put IT to use that never would have been considered just a year earlier. My only concern about the estimates I’ve provided here is that they are conservative from a growth perspective. Tighten your hats and put your local CSP and colocation provider on speed dial because the next five years are going to make the last five pale in comparison. A few examples of growth that were difficult, at best to predict:

    • Number of tower servers to replace demand for minis or mainframes. Within 10 years of the introduction of the IBM Clone server there were more than 10X in use globally than there ever were mainframes.
    • When blade servers matured, they offered anywhere from a 4-to-1 to a 16-to-1 space improvement over a typical 4RU pizza box server and yet server sales continued to increase, even though each of these blade servers had much more performance than the larger servers they were retiring.
    • Virtualization began really hitting its stride with VMware in 2007. The assumption for many years was a conservative average of 8-to-1 virtual machines to physical servers and yet, server sales continued unabated. Globally we have three times the number of physical servers now than we did in 2007.

Permissões de Postagem

  • Você não pode iniciar novos tópicos
  • Você não pode enviar respostas
  • Você não pode enviar anexos
  • Você não pode editar suas mensagens