Cloud computing and cloud adoption are perennial topics when talking about scalability and growth, but many enterprises still operate a significant portion of their workloads in legacy environments. With so much information on the reduced infrastructure costs and the elasticity of public cloud, why do organizations still do all the work themselves?
In this discussion with Sr. Product Marketing Manager William Chia, we talk about the challenges traditional IT teams face, the barriers to cloud adoption, and strategies to consider for making the leap.
Why organizations use traditional IT
The reasons that an organization may want to manage their own infrastructure are myriad and geared toward unique needs and/or limitations within their organization.
In highly-regulated industries such as banking and healthcare, or even government entities, there may be compliance concerns or risks that prevent them from utilizing public cloud. More control means more oversight and more accountability. "If I need to keep patient data private to comply with HIPAA, for example, if I keep 100% control of my systems and infrastructure I can ensure I comply. If I outsource to cloud services then I have to take different steps to ensure I'm not leaking PII," says William. Even though the big cloud providers – namely GCP, AWS, and Azure – have compliance built-in, some organizations may still be hesitant to have them assume those risks.
Protecting sensitive data
IT leaders surveyed in a Cloud Security Alliance report expressed that, while they are confident in cloud security capabilities, there are things that can go wrong beyond their control: Inside threats, compromised accounts, and misconfigured security settings up the stack that can all lead to security breaches. According to nearly 68% of the IT leaders surveyed, the ability to enforce corporate security policies is the number one barrier to moving applications to the cloud. "The top-level concern basically comes down to control and data privacy," says William.
For companies operating at a small scale, cloud computing’s pay-per-use model will almost always be cheaper than managing your own data centers, but for larger-scale organizations that isn’t always the case. "There's a breaking point… If you run on-prem, it actually could be cheaper than your cloud bill at huge scale, but you’re running so much software you’re basically running your own private cloud at that point," says William. For a long-term strategy, organizations have to weigh their CapEx vs OpEx costs, and while CapEx involves a large upfront expense in whole systems and servers, and the continued cost of maintenance, the computing volume could make this a worthwhile investment.
Another reason that companies may run their own infrastructure is because that’s how they’ve always done it. While not a very scientific answer, it’s the reality for many companies, especially those that grew before the age of cloud.
"Once upon a time, if you were a large enterprise and you had to run a lot of software, you had no choice but to manage it all yourself. And so now you have all these servers, you have all of these staff, and you have all of these business processes. You have a great deal of both physical and logical infrastructure and if you want to move to the cloud you have to change all of it. That comes at a very high cost," says William.
In the past, moving small amounts of data was relatively easy. When we start talking about exabytes of data, rather than terabytes of data, the process of migration becomes herculean. According to Jean-Luc Valente, the VP for product management in the cloud platforms and solutions group at Cisco, egressing that kind of data to a public cloud could cost as much as $30 million dollars.
The challenges of on-prem infrastructure
While organizations may have specific reasons for running on-premises infrastructure, that decision comes with distinct challenges.
Range of expertise
"Above a certain level, you are managing all of your infrastructure and you're managing all of your uptime. That's a lot of expertise. You need to become as good at operating a cloud infrastructure as Amazon or Google is, which is why those public clouds are so radically popular. In order to get there requires a lot of resources," says William.
Managing software and hardware
In order to manage uptime and security, operations teams need to perform software maintenance like upgrades and patches in addition to managing physical assets like servers, racks, power supplies, and network switches. At a certain point, an organization is devoting a lot of resources to just keeping things running rather than innovating, so all of these resources are being invested in undifferentiated engineering.
If it is not a core competency for your organization, then it’s undifferentiated engineering. "If you don't need to manage that on-premises data center or servers for a specific reason, then the cloud is more attractive because that's a high cost," says William. "You're spending a lot of engineering dollars on things that are not differentiating you in the marketplace."
Strategies for shifting to cloud
The benefits of "lift-and-shift"
In previous posts, we’ve talked about legacy and monolithic applications acting as a barrier for cloud adoption, but there can be some benefit to lifting and shifting those applications to the cloud. While you may not be able to take full advantage of microservices and cloud native application development, shifting those applications to the cloud does provide the benefit of reducing your operational overhead. This can provide an opportunity to learn new competencies.
"There's a separate set of competencies that you need to acquire to start running in the cloud. You don’t need to learn everything all at once. If you take a monolithic, on-premises app, simply lift-and-shift it into a VM in the cloud, that allows you to start to understand things like cloud billing, and gain some of the competencies of a cloud deployment pattern," says William.
Many organizations have opted to use both private and public cloud for a hybrid cloud infrastructure. These hybrid clouds blend the control and security of a private cloud, but also the flexibility and agility of public cloud. During periods of high usage, organizations can leverage public cloud’s pay-per-use model and save themselves from needing additional infrastructure. Organizations can use their private cloud for sensitive data and public cloud for developing and testing new applications. Having a hybrid cloud environment allows teams to manage their on-premises infrastructure and take advantage of public cloud scale.
While cloud adoption is widespread, many organizations have unique reasons to stay or migrate to an on-premises infrastructure. Cost, control, and risk mitigation continue to be the main drivers of on-prem vs. cloud decisions. Public cloud’s pay-per-use model may not be more cost effective for organizations that operate at higher scale, but a hybrid cloud model can offer organizations the flexibility to use public cloud during periods of high usage without having to invest in additional infrastructure. Both on-prem and cloud require unique and extensive operational competencies, so teams will need leaders that are skilled in these areas when making the switch.