When you make decisions about extending your infrastructure footprint into the cloud, you do so very intentionally. You hunt down analyst reports, ask peers for recommendations, and seek out quantitative research to compare the seemingly endless array of cloud-based options. But how can you be sure that you\u2019re getting the most relevant information for your business case? Bias exists and definitions matter. So each perspective is really just a single input in the decision-making process.<\/p>\n
The best process for evaluating any cloud solution involves four simple steps:<\/p>\n
The first step in approaching cloud adoption is to understand the resources your business actually needs. Are you looking to supplement your on-premises infrastructure with raw compute and storage power? Do your developers just need runtimes and turnkey services? Would you prefer infrastructure-abstracted software functionality?<\/p>\n
In the past, your answers to those questions may send you to three different cloud providers, but the times are changing. The lines between \u201cInfrastructure as a Service,\u201d \u201cPlatform as a Service,\u201d and \u201cSoftware as a Service\u201d have blurred, and many cloud providers are delivering those offerings side-by-side. While SoftLayer cloud resources would be considered \u201cinfrastructure,\u201d SoftLayer is only part of the broader IBM Cloud story.<\/p>\n
Within the IBM Cloud portfolio, customers find IaaS, PaaS, and SaaS solutions to meet their unique workload demands. From an infrastructure perspective alone, IBM Cloud offers cloud servers and storage from SoftLayer; containers, databases, deployment, and monitoring tools within Bluemix; and turnkey OpenStack private cloud environments from Blue Box. We are integrating every component of the IBM Cloud portfolio into a seamless user experience so that when a customer needs to add cognitive capabilities or a private cloud or video services to their bare metal server infrastructure, the process is quick and easy.<\/p>\n
Any evaluation of SoftLayer as a cloud provider would be shortsighted if it doesn\u2019t take into account the full context of how IBM Cloud is bringing together multiple unique, highly differentiated offerings to provide a dynamic, full-featured portfolio of tools and services in the cloud. And as you determine what you need in the cloud, you should look for a provider that enables the same kind of cross-functional flexibility so that you don\u2019t end up splintering your IT environment across multiple providers.<\/p>\n
Let\u2019s assume that you\u2019re primarily interested in deploying raw compute infrastructure in the cloud, since that\u2019s SoftLayer\u2019s primary focus. The seemingly simple first step in choosing the cloud infrastructure that best meets your needs is to define what \u201ccloud infrastructure\u201d actually means for your business.<\/p>\n
Technology analyst firm Gartner defines cloud IaaS as \u201ca standardized, highly automated offering, where compute resources, complemented by storage and networking capabilities, are owned by a service provider and offered to the customer on demand. The resources are scalable and elastic in near real time, and metered by use.\u201d While that definition seems broad, its Magic Quadrant for Cloud Infrastructure as a Service explains that when cloud resources are provisioned in \u201cnear real time,\u201d that means they must be deployed in minutes (not hours). To be considered \u201cmetered by use,\u201d they must be charged by the minute or hour (rather than by the month).<\/p>\n
Given Gartner\u2019s interpretation of \u201creal time\u201d and the \u201cby use\u201d measurement, bare metal servers that are fully configured by the customer and provisioned into a cloud provider\u2019s data center (usually in about two hours and billed by the month) aren\u2019t classified as cloud infrastructure as a service. That distinction is important, because many customers looking to extend workloads into the cloud are more interested in the performance of the resources than they are in provisioning times, and bare metal servers deliver better, more consistent performance than their virtualized counterparts.<\/p>\n
The performance angle is important. Many of cloud customers need servers capable of processing large, big data workloads (data mining, numerical and seismic analysis, processing and rendering 3D video, real-time social media analysis, etc.). These types of workloads generally consist of petabytes of data, and bare metal servers are better suited for running them\u2014and options like adding GPU cards for high performance computing make them even more enticing. The fact is that most virtualized cloud servers that can be delivered in minutes or less are not capable of handling these types of demanding workloads at all, or at least not as well as, more powerful bare metal servers that are available in just a couple of hours.<\/p>\n
In contrast to Gartner\u2019s definition, other analysts support the inclusion of monthly bare metal servers in cloud infrastructure decisions. In \u201cThe Truth About Price-Performance,\u201d Frost & Sullivan explains, \u201cBare metal servers provide the highest levels of raw \u2018throughput\u2019 for high-performance workloads, as well as flexibility to configure storage and network resources.\u201d And Forrester Research published a full report to address the question, \u201cIs bare metal really \u2018cloud\u2019?\u201d<\/a> The answer was, again, a resounding yes.<\/p>\n Using Gartner\u2019s definition, the majority of SoftLayer\u2019s cloud infrastructure as a service offerings are considered \u201cnoncloud,\u201d so they are not considered or measured in evaluations like the Magic Quadrant for Cloud IaaS. And without the majority of our business represented, the interpretation of those results may be confusing.<\/p>\n In practice, customers actually choose SoftLayer because of the availability of the offerings that Gartner considers to be \u201cnoncloud.\u201d For example, Clicktale, a SoftLayer client, explains, \u201cSoftLayer gives us the flexibility we need for demanding workloads. The amount of data we process is enormous, but SoftLayer\u2019s bare metal machines are the best out there and we have a high level of control over them\u2014it\u2019s like owning them ourselves.\u201d \u00a0<\/p>\n Our unique cloud platform with full support of both bare metal servers and virtual servers delivers compute resources that better suit our customers\u2019 workloads in the cloud. Whether or not you consider those resources \u201ccloud\u201d is up to you, but if you opt for a more limited definition, you\u2019ll cut out a large, important segment of the cloud market.<\/p>\n Once you settle on a definition of what meets your workload\u2019s needs in the cloud, it\u2019s important to evaluate how a given cloud resource will actually be used. Many of the factors that go into this evaluation are actually supplementary to the resource itself. Is it accessible via API? How can you connect it to your on-premises infrastructure? Will the data and workloads hosted on these resources be delivered quickly and consistently when your customers or internal teams need them?<\/p>\n While some of these questions are relatively easy to answer, others are nuanced. For example, SoftLayer’s data center footprint continues to expand around the world, but this seemingly pedestrian process of making servers available in a new facility or geography is only part of the story. Because every new SoftLayer data center is connected to a single global network backbone that streamlines and accelerates data transfer to, from, and between servers, as our data center footprint grows, our network performance improves to and from users in that geography to SoftLayer customer servers in every other data center around the world.<\/p>\n And what does that underlying network architecture mean in practice? Well, we\u2019ve run public network performance tests that show consistent results between 35 percent to 700 percent faster network speeds when compared to other \u201cleaders\u201d in the cloud space. Most industry reports, including Gartner\u2019s Magic Quadrant for Cloud Infrastructure as a Service, fail to acknowledge the importance of network performance in their assessments of cloud resources, focusing instead on the features and functionality of a given offering on its own.<\/p>\n The underlying platform capabilities and network infrastructure that support a given cloud resource aren\u2019t obvious when comparing the speeds and feeds of cloud server specifications. So as you evaluate a cloud provider, it\u2019s important to look beyond \u201cwhat\u2019s in the box\u201d to how cloud resources will actually perform, both on the server and between the server and your data\u2019s users. And the best way to get an understanding of that performance is to run your own tests.<\/p>\n The process of choosing a cloud provider or adopting a specific cloud resource cannot be purely academic. The nature of cloud computing allows for on-demand deployment of resources for real-world testing at a low cost with no long-term commitments. Making a decision to go with a given cloud provider or resource based on what anyone<\/em> says\u2014be it Gartner\u2019s MQ, Forrester, Frost & Sullivan, SoftLayer, or your nephew\u2014could have huge implications on your business.<\/p>\n SoftLayer will continue working with third-party research firms to demonstrate how our cloud infrastructure delivers up to 440 percent better performance for the cost<\/a> compared with our competitors, but those stats are meant to start a conversation, not end it.<\/p>\nUnderstand How You\u2019ll Use It<\/h3>\n
Test It Yourself<\/h3>\n