Best practices for balancing cost, security and performance in the cloud
Though 70 percent of IT leaders are investing in cloud, transitioning to a cloud environment is neither a quick nor easy endeavor. When embarking on this journey to cloud, the question IT leaders should ask themselves is how to best prioritize the undertaking.
There are three critical factors necessary for a successful transition: security, economics, and performance. The challenge is knowing how to balance each of these. In order to successfully transition and manage the cloud, it is important for businesses to determine the balance they need between these competing priorities.
Optimizing, cost, and benefits in the cloud
The hybrid cloud deployment environment, at its core, is driven by economics and more importantly, by the agility that virtualization brings. Private cloud or on-premise virtualized environments allow the organization to not compromise on the security and latency needs of sensitive applications and data.
Smaller organizations are able to fully go with public cloud until they reach a growth point where the operating cost of the public cloud becomes prohibitive. Once this point is reached, an application portfolio rationalization exercise becomes necessary to determine what stays in the public cloud and what needs to be moved on-premise i.e. a hybrid cloud approach.
One set of criteria to determine what needs to move on-premise includes latency SLAs, data security perimeters, and the mission-criticality of the application. If any of these are compromised, it could affect revenue, reputation and customer retention.
This possibility of eventual migration off the public cloud platform means architecture and technology choices have to be made that do not tie one so close to the public cloud vendor that this migration becomes impossible at a later stage.
A loosely coupled architecture where the technologies being used aren’t solely provided by the cloud vendor then becomes important. “Shadow IT” teams, which are essentially line of business local IT teams, sometimes prefer to (for sake of agility) make choices that might not fit with this line of thought, and this needs to be managed so that an organization doesn’t get held up when these non-standardized applications need to be brought in-house.
The usual suspect, in this case, is the database being used. Cloud vendors provide data management services that closely tie the application to their environment and make the migration impossible.
Protecting against vulnerability in cloud security
Historically, some enterprises have been reluctant to migrate to the cloud due to privacy and security concerns. But these are mostly alleviated by public cloud vendors getting their data centers compliance certified for the likes of HIPAA and PCI compliance. But high risk-exposure applications such as those in the financial services industry are unlikely to use a public cloud due to the fact that security is managed by a third-party host and, a hack into the third party systems makes the organizational data vulnerable.
A key benefit of the cloud is scalability, allowing increased agility to pull data that needs to be processed and analyzed in real-time. However, the more data you are managing, the more likely you are to sacrifice provisioning efficiencies and open more vulnerabilities for a breach. It is critical that security measures are embedded in your database in the cloud and are continuously adjusted as you scale and grow your data pool.
Maintaining speed and agility in the cloud at scale
When considering performance, typically the layer that processes data is the largest contributor in the overall latency equation. When designing real-time applications that require the lowest end-to-end latency, it is desired to choose a data storage and processing technology that supports that objective and also maintain that performance in both public and private cloud environments.
Modern data requirements go beyond just the high-performance data ingestion or mutation. It has evolved to the stage where in-line decisions need to be made as the data is being generated and ingested. A cloud-ready data layer will have both the storage and the processing capabilities.
In today’s world, ever-evolving intelligence is being created from machine learning. The data storage and processing layer now requires the additional capability to be able to employ these machine learning models into their decision making processes. This moves the insights generated into becoming a foresight in transactional decisions. Examples of these types of applications are credit card fraud prevention, telecom fraud prevention, portfolio risk analysis and hyper-personalization.
In order to alleviate concerns and experience the most successful transition when migrating to the cloud environment, there must be balance. Businesses must simultaneously manage cost, security and performance in order to optimize the best of each priority in the cloud. These priorities must be addressed with clear definition of roles and responsibilities of applications. Further classification of these applications will determine which of them are insight generation and which are insight application.
This application classification helps manage cost and performance by subdividing them into stateless compute and stateful data layers in the cloud. Prioritizing certain workloads into multiple private clouds distributes cloud fees to make them easier to manage and provides companies with lower latency on their data through a wider range of infrastructure.
Real-time decisioning is still better served on-premises due to stringent latency service level agreements (SLAs) and highest performance guarantees that cannot be held hostage to the provisioning efficacies of the public cloud.
Now, this on-premise computing can be virtualized into private clouds and incorporated into a multi-cloud environment. That being said, cloud diversity also poses extra risks and potential loss of control of cost. Managing this diversity is a demanding and time consuming task. One solution to this is finding a data platform that has a storage layer, stateful data processing layer and also a machine learning model deployment environment that can operate in both public and private cloud environments.
Take the next steps in your cloud strategy. See how CenturyLink can help.
This blog is provided for informational purposes only and may require additional research and substantiation by the end user. In addition, the information is provided “as is” without any warranty or condition of any kind, either express or implied. Use of this information is at the end user’s own risk. CenturyLink does not warrant that the information will meet the end user’s requirements or that the implementation or usage of this information will result in the desired outcome of the end user.