The 4 most common mistakes when moving to the cloud

1. The connection to the Internet / cloud is put on the back burner

There are many things to consider when migrating to the cloud. This starts with the internet connection. An existing line is usually available, but is it fail-safe enough, does it have sufficient bandwidth and a powerful backup? The performance of your cloud application is only as good as your network! For example, when a professional services company was nearing the end of a cloud migration project, the team realized that the public internet would not be suitable as an access route to the cloud, as business-critical systems require a much faster and more stable line. The client had planned to call their provider and purchase another fiber connection. However, the provider had neither the fiber in the ground nor the capacity to the cloud entry point, and the lead times were several months. As a result, the project had to be postponed, resulting in unnecessary additional costs and extra work. Therefore, involve your provider at an early stage to ensure a suitable connection to the cloud. Make sure that your provider can provide the service and agree clear SLAs, which should include, among other things:

• Guaranteed throughput: “Best effort” delivery is not an option.
• Low latency: This guarantees performance and a positive user experience.
• Privacy and security: Check whether you need/want a completely private connection to your cloud provider.
• Reliability and resilience: There should be several resilience options. What level of availability can your provider guarantee?
• Flexibility: Consider planned and unplanned workloads. Can you adjust your bandwidth so that your network offers the same scalability as the cloud?

2. Insufficient planning for the future

Until now, capacity planning has been based on gradual increases in IT workloads and network capacity. As the company grows and the demand for bandwidth-intensive applications increases, more bandwidth will be required each year and this will then be increased gradually. Such an approach is no longer recommended for three reasons:

• Expensive: If you plan a gradual increase in workload and network capacity, you will not be able to make the best use of your budget. You have to plan for maximum demand from the outset and spend too much on redundant network capacity.
• Fluctuating demand for the cloud: Modern businesses have more and more unplanned workloads and departments with their own priorities and initiatives. If you don’t build a scalable network, unplanned traffic can impact your entire business.
• You cannot use innovations quickly: In the cloud, it is much easier to update or introduce new services using technologies such as artificial intelligence (AI) and blockchain. These can automate key processes, manage performance without human intervention and speed up transactions, but all require additional bandwidth. Your network needs to be able to meet the demands of future technologies, you shouldn’t have to wait 90 days for capacity changes, or limit day-to-day operations by reaching network capacity.

We therefore recommend proactively planning for innovations. Ask your provider what flexibility options they can offer you. Rather than overbuying capacity to meet potential demand, find a provider that can scale the network as needed. A flexible network will allow you to handle unplanned traffic spikes and take advantage of new market opportunities or technologies.

Ask your provider these questions:

• How quickly can I scale?
• How can you support the performance requirements of my apps?
• What are the implications of adding these apps to my cloud stack?
• What do you suggest for disaster recovery?
• What security levels do you offer?
• How easy is it to switch from one cloud to another?

3. No consideration of the complex app landscape

According to a 2019 study by McAfee, the average large company uses over 1,900 cloud services. However, IT managers estimated that they only use around 30. An example: A company wanted to replace physical servers by migrating to the cloud. They planned the project assuming they had 700 physical servers and discovered they had 1,100 at the start of the consolidation process. To make matters worse, by the end of the 18-month consolidation project, instead of 1,100, they were running more than 2,000 virtual machines on 1,300 physical servers. This lack of understanding of the existing app landscape meant that the cloud migration project, which was intended to consolidate the infrastructure, only served to complicate the latter and did not achieve the desired efficiency savings. Network capability and app performance are closely linked. Depending on the application, access via the public Internet is sometimes sufficient, but for business-critical applications (such as ERP, payment processes, CRM systems) you may need a dedicated line directly to the data center with fixed bandwidths, SLAs, private IPs or ultra-low latency.

Cloud application checklist:

• Identify business-critical applications (telephony, ERP, payment processes…)
• Decide what should remain on-premise and what should be migrated
• Define performance and availability requirements
• Carry out a BYOD assessment of your company’s shadow IT and the resources consumed
• Evaluate cyber security vulnerabilities of your network
• Record documented problems, individual sources of error and risks
• Consider data and file access, resilience and security
• Carry out an audit to plan the network capacity
• Design a roadmap for capacity expansions and geographical expansions in the future

4. Security and availability of the cloud are not taken into account

In a survey of 983 IT decision-makers, 77% saw improving security as a top priority. A major advantage of the cloud is that hyperscale providers, for example, can offer security features that are not available to most organizations. For most companies moving important data such as customer information and employee data to the cloud, it is important to pay attention to its security to avoid costly financial and reputational damage. Example: At an international bank, a hacker accessed over 100 million customer records, including addresses and credit rating data stored in the cloud. Although no credit card data was leaked, the bank was exposed to significant reputational damage as a result of unfavorable media coverage. Such incidents can also lead to legal action in EU and international courts if they constitute a breach of customer data protection laws such as the GDPR. Never take cybersecurity in the cloud for granted. Your cloud vendor, provider and company share responsibility for cybersecurity. Consider the different levels of network security. There are three levels of security if you want to connect your premises to the cloud:

• Public Internet: Like a bus, you have to share with many other passengers who may have malicious intentions. A bus is also at risk of getting stuck in traffic.
• Cloud VPN: Creates a tunnel over the internet that protects you from the public. This is like having a private cab while still driving on the public highway. You can still get stuck in traffic or rerouted, but no one can see what data you have on your journey.
• Partner Interconnect: The best connectivity option for companies who want to ensure that their data reaches the cloud on time and is not intercepted. Here you are driven in your own limousine on your own private road.