A 2020 survey found that at least half of data center workloads will remain in enterprise data centers until at least 2022. Businesses are hesitant to move more mission-critical operations to the cloud until there is more openness, transparency, and accountability. The following participants in the Evolving Data Center Networks session of ONUG Fall 2020 would agree:
The session began with each participant identifying one area of concern when moving to the cloud.
For Abby Kearns, the stumbling block was changed. Processes do not work the same in a cloud environment as they do in an on-premise enterprise, but people struggle to make the shift. Most people have difficulty letting go of old processes. Rather than redesigning a process to fit the environment, people try to adjust the environment to fit the process.
Carlos Matos’ stumbling block was the perception that the cloud was “Nirvana.” People have high expectations for a cloud implementation without understanding that the move cannot happen all at once. Implementation must be a process that moves “old school” to “new school” thinking.
Charles Greenaway sees the pandemic as having the potential to both accelerate and depress cloud implementation. Because of COVID, organizations have turned to cloud-based solutions to enable employees to work remotely and deliver solutions to customers. At the same time, companies are hurting financially, putting a drag on their willingness to spend money.
Parantap Lahiri’s concern was the lack of understanding of the underlay network by those using the system. Although applications should determine what the underlay needs to provide, the system’s complexities should decide what needs automation.
Since most businesses will continue to operate in a hybrid environment for at least another two years, network automation must be addressed for both cloud and on-prem installation. Higher demand for network consumption challenges an enterprises’ ability to scale its infrastructures, whether on-premises or in the cloud. The struggle has expanded to include network teams who must scale while managing cloud technologies alongside data center networks.
IT personnel have to learn a different way of managing cloud-based network services. Although networking is similar in function to on-prem infrastructures, they require a different approach to provisioning, managing, and troubleshooting. Those differences create a learning curve that many organizations cannot absorb. Experience with physical hardware simply does not apply in the cloud.
The best solutions are ones that integrate on-prem and cloud network management into a single system or at least one that is easy to traverse when moving from one network to another. That means looking at each environment to determine the best automation tools to deploy and finding ways to merge that data into useful information.
How does an organization choose what to move to the cloud, especially when it comes to network automation? Legacy systems that reside in on-prem data centers have existed long enough that they have tools with a complement of bells and whistles. Many of these extras will not transfer to the cloud. That means doing without these custom tools or replacing them with cloud-compatible ones.
All participants agreed that the traditional ways of programming are incompatible with the cloud and that the situation is further complicated by the lack of standards for cloud providers. Carlos believes vendors need to help by establishing a framework for reporting status. The current use of APIs is not enough. With better reporting, enterprises could create solutions that would allow for true transformation.
The ability to design networks that allow for dynamic separation of resources will add to network security, making it faster to isolate potential threats. Security is a shared responsibility, but a standardized framework across all cloud providers would facilitate cloud migration. At this point, some mission-critical solutions cannot be moved unless companies can achieve more control.
As Charles pointed out, systems must be “cloud agnostic.” At present, organizations with multi-cloud implementations must develop cloud-specific programs for processing data sent from the provider. Such development is costly, but no business can afford to be tied to a single cloud provider.
A shared concern was the lack of understanding of the network underlay of many IT professionals. They have never been in a physical data center or saw how the physical components interconnect to provide system resources. The participants weren’t suggesting that every IT person needed to run wire but that a better picture of the physical layer would be helpful.
More complete awareness of the physical components would help programmers understand that the infrastructure of the cloud and the on-premise data centers are drastically different—old methods of programming place a drag on the underlay. For companies to scale, they need everyone to be effectively and efficiently using the available resources.
For data centers to evolve into cloud-native enterprises, cloud-providers have to change their solutions to give companies the transparency they need to maintain a level of control over their digital assets. Until providers are willing to deliver that functionality, corporations will continue to keep mission-critical resources in data centers. With compliance regulations in place for many industries, companies cannot afford to relinquish control to a third-party.
Internally, organizations have to change the IT mindset to embrace the differences between a cloud-based and enterprise-based environment. Some businesses have started to apply cloud approaches to their on-premise data centers, making it simpler for teams to manage a mixed environment. For the immediate future, most data centers will operate a hybrid infrastructure.
To learn more about how data centers are moving to cloud-native operations, join us for ONUG Spring 2021, May 5-6.