Five Tips For A Successful Cloud Migration
Almost immediately ready for use and without any operational effort – moving to the clouds is not that easy. You should consider five best practices to test applications and IT environments’ “cloud suitability” in a proof-of-concept (POC). With the Proof-of-Concept ( PoC ), IT managers can test best practices and localize risks on the rocky road to the cloud before they pack their backpacks with applications and finally head toward the summit. The problem: Many PoCs run out of breath after just a short distance. Cloud initiatives remain half-finished, encounter insurmountable technical problems or fail because of cost and resource issues. Â
At the end of the long-planned excursion, only a few smaller experiments are running in the cloud while the rest of the IT infrastructure stays on the ground. Flexera’s 2022 State of the Cloud Report states that cloud migration is a challenge for 73 percent of organizations. The application dependency mapping (53%), the technical feasibility check (48%), and the cost comparison of cloud and on-premises (41%) are particularly problematic. Anyone who takes their first steps from the data center is therefore well advised to take five essential tips into account.
Spoiled For Choice (Application)
The cloud is not suitable for every application. However, every company has applications in its IT portfolio that are worth moving. When it comes to picking out an IT asset for a cloud test run, PoC teams can be very picky. The application should not be artificially constructed but should address a real business or technological need within the day-to-day business operations. It is better to stay away from highly complex, monolithic applications and concentrate on a project that will most likely enable a migration in 30 to 60 days. Caution should also be exercised for exceptional security and compliance requirements. The cloud offers rapid provisioning, scalability, and greater agility when processing workloads.Â
The selected application should match this profile. Good test balloons are, e.g., B. mobile apps and applications related to social media, big data, or batch processes. A completely new solution offers more leeway than an existing application in the data center because it can be designed from the start to run in the cloud and does not have to meet current requirements. Ideally, after the PoC, the company will gain a new IT asset that generates real ROI. Nevertheless, developing a new application is not necessarily the focus of cloud testing. It is primarily about the learning effect and improving the cloud migration process. The basic rule is: to start gradually, build on successes, and slowly but surely increase the complexity.
Copy-Paste Does Not Work.
Even if it is tempting: There is no copy-paste button for cloud migration to replicate the selected application 1:1 in the cloud. Virtual machines (VM) are not cloud instances and cannot be easily moved. Typically, the IT landscape in companies is built from the bottom up: it starts with the hardware, followed by the installation of hypervisors and operating systems, and ends with the deployment of applications. A successful cloud implementation turns this approach on its head.
The focus is initially on the application, and only the second step asks what requirements the cloud infrastructure must offer to run it. A business service mapping is necessary to clarify whether the application selected for the PoC is suitable for the cloud and what affects the migration has on other assets in the IT landscape. An automated bottom-up approach maps dependencies holistically and creates transparency. The IT visibility gained in this way reveals how the PoC application to be migrated is connected to other applications and which dependencies have to be taken into account.
Public Vs. Private: Where Does The Starting Gun Go?
Once the application has been determined, the next step is to determine the cloud. Companies have the choice between public cloud (e.g., Amazon Web Services (AWS), Google Cloud Platform (GCP), IBM Cloud, Oracle Cloud, and Microsoft Azure) and private cloud (e.g., based on OpenStack, CloudStack, and VMware vSphere ). Even if the private cloud is the declared end goal, it is worth starting with the PoC in the much more agile public cloud. Public clouds often use the same hypervisors and orchestration technologies as private clouds. This way, the PoC team can quickly and easily familiarize themselves with cloud technology without investing much time and money. If the application runs in the public cloud, it can be deployed in the private cloud.
For example, private clouds are a good choice for applications that require high computing power. While public clouds offer only a limited number of hardware options, companies can deploy as many cores and performance as their budget allows in the private cloud. Security and compliance – e.g., B. when it comes to safety-critical systems and applications (KRITIS) – are also good reasons to provide applications via the private cloud.
The Crux Of The Technical Requirements
As soon as the PoC application is in the cloud, optimization begins. On the technical side, there are many hurdles to overcome. Experience has shown that the biggest headache is the storage performance, which often cannot keep up with the usual power of your storage area network (SAN) at the beginning. Here it is worth looking at alternative cloud providers that offer adjustable IOPS performance or provide a combination of bare metal and cloud servers. Combining bare metal servers for high-performance storage and cloud servers for elastic workloads can significantly increase performance . In data centers, companies often take up storage space multiple times and install the same volume on multiple servers simultaneously. However, cloud storage is not always possible.
In some cases, companies can circumvent these limitations, e.g., B. Use file systems like GlusterFS to share storage space in the cloud. Performance, licensing, and the associated costs must also be considered. For example, is the PoC application sized by physical CPUcores or hosts? Or is it based on a usage-based model? The test should clarify which cloud provider offers the best conditions and which discount campaigns can be used to reduce costs further. Those with specific security and compliance requirements within the PoC should consider disk encryption and a Virtual Private Network (VPN) to protect data at rest and in transit.
Measure, Analyze, And Optimize
A PoC only makes sense if critical figures are available at the end to assess the migration project’s success objectively. The goals associated with the cloud PoC can vary widely. However, there are still some basic parameters. Does this include the question of business agility: how time-consuming and costly was the migration? And what are the benefits of the new cloud application for employees? For example, can business requirements and opportunities now be managed more quickly and flexibly? Or can IT now offer new services in additional regions at short notice? The technological added value indicates the ROI that is possible with the new application in the cloud. However, the cost is the most precise way to measure the success of the cloud application.
The public cloud offers a high level of transparency and enables a comparison between cloud and on-premises costs. The savings that result from the lower maintenance and administration costs in the cloud also belong in the cost calculation. the analysis of the PoC results also reveals further potential for optimization in the cloud. Perhaps teams will find new starting points to automate specific routine tasks. It may be time to consider implementing a self-service portal to provide business users and developers with easy access to cloud resources. Or the test uncovers gaps in the technical assessment of business services or within the IT asset data. This does not change the success of the cloud PoC – at least not if you understand it as an important lesson learned on the way to the cloud, regardless of the result.
Read Also: The Chatbot, The Virtual Agent, And The Human Service
Share this content:
Post Comment