What Is Infrastructure-as-a-Service?
The business cloud is ever-expanding and evolving. Early cloud platforms like Microsoft Office 365 and Google G Suite are widely used by organizations of all types and sizes and have handily replaced boxed software with more versatile and collaborative cloud-enabled solutions. Similarly, specialized applications such as Adobe Creative Cloud, NetSuite, and many others are thriving on the cloud as OS-agnostic options accessible to users on a per need basis in all-inclusive subscription packages.
Salesforce is probably the first cloud app that many companies used, but there are now others that your organization probably uses every day, ranging from FedEx.com and UPS.com to conferencing apps such as GoToMeeting and Cisco WebEx Meetings. If it seems like many, if not most, of the apps you use to reach outside your company’s walls are cloud-based apps, then you’d be right as they are everywhere.
Meanwhile, companies are replacing their own infrastructure with cloud services as well. In some cases, it’s simply for storage and backup where Box (for Business) and Dropbox are everywhere, while others, such as Carbonite Cloud Backup and MozyPro for Business, are quietly providing backup for organizations from the Fortune 500 to your home office. Companies have also started using cloud services to offload compute services that once ran in their own data center by using what we now call Infrastructure-as-a-Service (IaaS).
Software-as-a-Service (SaaS), which was the first type of cloud service, has become a de facto standard for business. It’s become so standardized that, in some cases, there is no standalone software choice available. When’s the last time you saw the standalone FedEx app on a computer? In other cases, they have brought new capabilities to businesses that simply didn’t exist before, such as off-site storage with Box (for Business) and Dropbox. And while video conferencing apps do still exist, they are on their way out, except in the form of room-based systems in big organizations and government agencies.
Infrastructure services in the cloud are now becoming a standard for organizations of all sizes. Large enterprises are well along in making the move to IaaS, and small to midsize businesses (SMBs) are now moving as well. Right now, the prime reason more companies haven’t moved is their lack of technical talent to move apps to the cloud and implement the apps there. But the growing ranks of consultants and professional services are now moving that possibility into the reach of smaller organizations.
For most organizations, a move to the cloud saves money when it comes to expanding a data center or implementing a new one. There’s very little capital expenditure involved with a cloud implementation and, while there are monthly charges for cloud operations, they’re almost always far less than paying an IT department to do the same work.
In addition to being less expensive, IaaS has operational advantages as well. If a piece of infrastructure in your data center goes down, then you’ll have to find a staff member to go fix the problem. This may mean calling someone in over a night, weekend, or holiday; it will mean extra pay, and it might also mean that you stay down if the item—whether it’s a server or a switch—needs a replacement that’s not available until the next business day.
That problem is reduced with cloud services. Depending on your service-level agreement (SLA), the cloud provider has the responsibility to make sure your workload stays up and running. While that doesn’t guarantee 100 percent uptime (after all, you still need an internet connection, and that depends on infrastructure in your shop and network connectivity that’s not controlled by your cloud provider), it’s a level of reliability that’s hard to match inside your own data center.
Cloud-based infrastructure is also more flexible. You can add or shrink capacity as your needs change or as demands change because of seasonal variations. You can stay with your preferred server operating system (OS) without demands that you upgrade on any schedule but your own.
You have access to top-notch data and physical security, too. While some security choices may add to the cost of your virtual servers, you get to choose exactly what type and level of security you need for your work. What’s also a plus is that physical security is provided as part of the deal, and it’s probably far better physical security than you can provide at your own facility.
In this IaaS solutions review roundup, we take a look at five of the largest publicly available cloud services, including Amazon Web Services (AWS), DigitalOcean, Google Cloud Platform, IBM Cloud, and Rackspace Managed Cloud. (Microsoft Azure, which participated in our previous IaaS solutions roundup, declined to participate this time around, even after we tried contacting them several times.)
All of these cloud services are available to business users that need a compute platform for general-purpose work. Of course, these cloud services provide a vast array of services beyond just compute platforms, so before you choose what we found, also make sure that your cloud provider can offer services that you think you might need in the future. While it’s possible to migrate from one cloud provider to another, it’s not necessarily easy.
IaaS Cloud Types
There are three types of cloud infrastructure. The type we’re reviewing in this review roundup are public clouds, which means that anyone can get access to cloud services, including virtual server platforms, along with a vast array of related services. Some services, such as AWS, have literally thousands of options of things you can do to customize your cloud.
There are also private clouds, which are servers using cloud technology that may exist in your data center, a colocation site, or even as a private cloud located within the physical infrastructure at a cloud provider. There you get dedicated servers and other hardware, but the infrastructure is maintained and operated by the cloud provider’s staff.
For many businesses, there’s a middle ground. A hybrid cloud uses a public cloud service for some functions and private cloud services for others. The reasons for this vary widely. Usually the private cloud service is either there because there’s something about the software that doesn’t lend itself to the public cloud; for example, if it requires a proprietary platform. Or there may be something about the data; for example, many organizations don’t want their most sensitive data to reside outside of their premises.
There is a third reason, which is performance. The public cloud necessarily brings some level of latency. Each switch or router introduces some small delay in a network connection, plus those pesky laws of physics limit how fast a packet of data can travel over a network. The only way to overcome the latency in cloud communications is to reduce the amount of infrastructure and reduce the physical distance. Keeping the cloud in the building solves both problems with latency.
It’s worth noting that there are several types of hybrid clouds. One simply uses different clouds for different functions. The other uses your virtualization software, such as VMware (6.00 Per User Per Month at VMware) , to manage data in both places. This takes some knowledge to set up, but it’s seamless once that’s done.
Choosing Your Cloud
We’re testing a series of public clouds in this IaaS solutions roundup, and we’re basing our tests on their ability to deliver a compute platform that will perform specific tasks. In addition, we’re testing performance by using Geekbench 4, a cross-platform benchmarking app that’s designed to treat all platforms the same, regardless of the OS.
Cloud services are more than just a sum of their different parts. Cloud vendors tend to have areas of specializations or seem to fit some tasks and business processes better than others. Each provider will feature additional features worth considering. This could be anything from development environments, security services, or third-party integrations. Some platform providers, like AWS offer a large magnitude of options which requires a professional consultant to serve as a guide to help nail down the features specific businesses need.
One thing to realize in choosing your cloud is that the hardware you’re going to end up with will have little to differentiate them. Basically, you’re going to find yourself choosing a compute platform based on one or more Intel processors, with an amount of memory that you specify, and some kind of mass storage, which may be hard disks or solid-state disks (SSDs). You will get to choose load balancers, OSes, backup methods, and a wealth of other characteristics.
But the differences between one cloud vendor an another aren’t that great; they’re all selling access to the same type of server. The differences lie elsewhere. For example, while every cloud vendor offers Linux, they’re not all of the same distribution of Linux. Most, but not all, offer Microsoft Windows Server, but not all offer the same versions. So, with one vendor your latest choice may be Windows Server 2016, and with another, your latest choice may be Windows Server 2012 R2. A few cloud vendors offer Free BSD Unix, some offer VMware, and some don’t. One cloud vendor, Rackspace, offers access to the other major cloud providers through its own cloud.
When choosing your cloud provider, it’s important to pay attention to the environment that is the focus of the cloud offerings. Microsoft Azure, for example, focuses on services offered by Microsoft, so while you can build servers using Linux, you can also build them with Windows Server and you have access to virtually anything Microsoft makes that will work in a server environment.
For its part, IBM Cloud focuses on providing that company’s all-encompassing service platform, and it includes features such as access to the IBM Watson ($360.00 Per User Per Year for the Plus Edition at Software Advice) supercomputer, along with a large collection of prewritten routines you can use to take advantage of IBM Watson. AWS, on the other hand, tries to be all things to all people, and the result can sometimes be bewildering.
One thing that I found out as I ran the tests on the basic cloud environments is that raw performance is not a differentiating factor. In addition, if you need to revise your performance, then these providers will provide a way to add processors and storage or to reduce those when you don’t need them.
A more important factor is to talk to vendors in which you’re interested to see what SLA guarantees they make regarding uptime—and what they do if they don’t make their uptime promises. For example, if your server goes down for an hour, then do you get a refund on the time or does the provider promise to move your workload to a new server? And if the latter’s the case, then how long will it take? If you’re depending on a cloud service, then downtime costs you real money, and a few dollars as a refund doesn’t help you stay in business.
So, you’ll want to examine providers’ SLAs carefully. You may want to get help when you do, either by calling in your corporate counsel or by calling in a consultant that specializes in cloud services. But because the potential for downtime is critical for your business, it is important that you learn exactly what’s promised, and whether there are ways you can tailor the providers’ SLAs to suit your needs.
Finally, you’ll have to determine the cost to you of the cloud configuration you think you’ll need. The cloud providers have pricing calculators, but the usefulness of those calculators is limited, to put it politely. The fact is, in many cases, it’s nearly impossible to know exactly what a cloud service is going to cost you. This is especially the case with AWS, which is so complex that, when we asked for help from Amazon to determine the price of our standard configuration, we never got a solid answer. In some cases, it will pay to hire a consultant to design the configuration for you and tell you what the price should be. It’s easy to get lost in the configuration options, and you may find it tricky to even get what you need, much less figure out the price.
Once you think you’ve figured out the price and the proper services, ask the cloud provider what they think it will cost and whether there are better configurations. Expertise is important, and you may have to pay to get it.
If the idea of a public cloud sounds intimidating, then that’s no surprise. But remember that the basic idea is fairly simple, and just because you’ve got a lot of options doesn’t mean you have to use them. Setting up and running a basic compute platform is quite simple with any of the tested platforms. The cloud providers have done a lot to make the basic tasks simply a matter of choosing from a menu.
What’s even better is that if you don’t need a specialized or complex capability, then you don’t have to pay for it. All of the cloud vendors charge by what you actually use, so you’re not paying for capability you don’t need. In addition, most of the cloud vendors have the option of trying out their service for free. This means you can see if the cloud is right for you before you have to pay anything or do anything permanent. Plus, you have a way to make sure that the cloud you’re getting is the cloud you need.