As IT pros aim to make the most efficient use of their budgets, there is a rapidly increasing range of infrastructure options at their disposal. Gartner’s prediction that public cloud spending in North America will increase from $2 billion (£1.2bn) in 2011 to $14 billion (£8.5bn) in 2016, and 451 Research’s expectation that co-location demand will outpace supply in most of the top 10 North American markets through 2014, are just two examples of the growing need for all types of outsourced IT infrastructure.
While public cloud services in particular have exploded in popularity, especially for organisations without the resources to operate their own data centres, a “one size fits all” myth has also emerged, suggesting that this is the most efficient and cost-effective option for all scenarios.
In reality, the public cloud may be the sexy new sports car — coveted for its horsepower and handling — but sometimes a hybrid model can be the more sensible approach, burning less gas and still getting you where you need to go. It all depends on what kind of trip you’re taking. Or — put in data centre terminology — the most effective approach depends on the type of application or workload, and is often a combination of infrastructure services, ranging from public, private, and “bare metal” cloud, to co-location and managed hosting, as well as in-house IT resources.
The myth of cloud fuel economy
Looking deeper into the myth of “cloud costs,” as part of a recent “Data Centre Services Landscape” report, Internap recently surveyed 100 IT decision makers to gain a cross-sectional view into their existing, current, and future use of IT infrastructure. Almost 65% of respondents said they are considering public cloud services, and 41% reported they are doing so in order to reduce costs.
But when you compare the “all-in” costs of operating thousands of servers over several years in a well-run corporate data centre, or collocating in a multi-tenant data centre, against the cost of attaining that same capacity on a pay-as-you-go basis via public cloud, the cloud service will lose out nearly every time.
The fact that co-location can be more cost-efficient than cloud often comes as a surprise to organisations, and is something of a dirty little secret within the IaaS industry. But for predictable workloads and core infrastructure that is “always on,” the public cloud is a more expensive option because the customer ultimately pays a premium for pay-as-you-go pricing and scalable capacity that they don’t need — similar to driving a gas-guzzling truck even when there’s nothing you need to tow.
Balancing the race car handling of cloud with the safety of a family hybrid
This is not to suggest that cloud is without its benefits. Public cloud makes a lot of sense for unpredictable workloads. Enterprises can leverage it to expand capacity on-demand without incurring capital expenditures on new servers. Workloads with variable demand and significant traffic peaks and valleys, such as holiday shopping spikes for online retailers or a software publisher rolling out a new product, are generally well-suited for public clouds because the customer doesn’t pay for computer capacity that they don’t need or use on a consistent basis.
One of the biggest benefits of cloud services is agility. This is where the cloud truly shines, providing accessibility and immediacy to the end-user. However, the need for a hybrid approach also arises here, when agility comes at the expense of security and control. For example, the agility versus control challenge is often played out, one such example being when a CIOs became upset when she found out that employees within most of the company’s business units were leveraging public cloud services without her knowledge. This was especially unsettling, given that she had just spent millions of dollars building two new corporate data centres that were only half full. Something had gone wrong here, and it’s related to agility.
A major contributing factor to the surprise popularity of public cloud services is the perceived lack of agility of internal IT organisations. For example, it’s not uncommon for it to take IT executives quite some time to turn up new servers in corporate data centres. And this isn’t necessarily the fault of IT, since there are a number of factors that can — and often do — present roadblocks, such as the need to seek budgetary approval, place orders, get various sign-offs, install the servers, and finally release the infrastructure to the appropriate business units (a process that can easily take several months). As a result, employees and business units often begin to side-step IT altogether and go straight to public cloud providers, corporate credit card in hand, in an effort to quickly address IT issues. The emergence of popular cloud-based applications make this scenario a common occurrence, and it illustrates perfectly how the promise of agility can end up pulling the business units toward the public cloud, at the risk of corporate security.
The CIO is then left scrambling to regain control, with users having bypassed many important processes that the enterprise spent years implementing. Unlike internal data centres or co-location environments, with a public cloud, enterprises have little to no insight into the servers, switches, and storage environment.
So while agility is clearly a big win for the cloud, security and control issues can complicate matters. Again, a hybrid, workload-centric approach can make sense. Use the cloud for workloads that aren't high security, and consider the economics of the workload in the decision as well. Some hybrid cloud solutions even allow enterprises to reap the agility benefits of the cloud in their own data centre, essentially an on premise private cloud.
As businesses continue to evolve, it will be critical to go beyond the industry’s cloud hype and instead build flexible, centrally-managed architectures that take a workload-centric approach, and apply the best infrastructure environment to the job at hand. Enterprises will find such a hybrid solution is usually of greater value than the sum of its individual parts.
Carpooling with “cloudy co-lo”
One area that has historically been left out of the hybridisation picture is co-location. While organisations can already access hybridised public and private and even “bare metal” cloud services today, co-location has always existed in a siloed environment, without the same levels of visibility, automation, and integration with other infrastructure that are often found in cloud and hosting services.
But these characteristics are likely to impact the way co-location services are managed and delivered in the future. Internap’s survey found strong interest in “cloudy co-lo” — co-location with cloud-like monitoring and management capabilities that provides remote visibility into the co-location environment and seamless hybridisation with cloud and other infrastructure, such as dedicated and managed hosting.
Specifically, a majority of respondents (57%) cited interest in hybrid IT environments; and, combined with 72% of respondents expressing interest in hybridising their co-location environment with other IT infrastructure services via an online portal, the results show strong emerging interest in data centre environments that can support hybrid use cases as well as unified monitoring and management via a “single pane of glass.”
Driving Toward a Flexible Future
A truly hybrid architecture — one that incorporates a full range of infrastructure types, from public and private cloud to dedicated and managed hosting, and even co-location — will provide organisations with valuable, holistic insight, and streamlined monitoring and management of all of their resources within the data centre, as well as consolidated billing.
For example, through a single dashboard, organisations could perform tasks, such as: remotely manage bandwidth, inventory, and power utilisation for their co-location environment; rapidly move a maturing application from dedicated hosting to co-location; turn cloud services up and down as needed, or move a cloud-based workload to custom hosting. Think of it as your hybrid’s in-car navigation system with touch-screen controls for everything from radio to air conditioning to your rear view camera.
The growing awareness of the potential benefits of hybridising IT infrastructure services reflects the onset of a shift in how cloud, hosting, and even co-location will be delivered in the future. The cloud model, with its self-service features, is one of the key drivers for this change, spurring interest among organisations in maximising visibility and efficiency of their entire data centre infrastructure ecosystem.
This article first appeared in the October 2013 issue of Cloud Computing Intelligence magazine – Subscribe now to get the latest cloud computing information delivered to your desk.
About the Writer
Adam Weissmuller is the Director of Cloud Solutions at Internap. In this role, Adam has recently launched the Internap cloud solution suite. Adam has more than 15 years of IT experience, holding roles in product management, operations, and sales. Adam is also a 10-year veteran of the hosting industry and, prior to joining Internap, held management positions at PEER 1 Hosting and Rackspace. He recently presented on “Overcoming Latency: The Achilles Heel of Cloud Computing” at Cloud Expo West.