Cloud Complexity - Bringing Resources together in Multi-cloud Environments

The world is still getting used to operating within the cloud. Moving to the cloud is challenging for many organizations. So why do we see a rise in the adoption of multicloud strategies? In this blog, we will explore why this trend is worth considering for your organization, as well as look at the challenges that it brings.

What is Multicloud?

Multicloud strategies emerge when a combination of private or public cloud providers (Azure, AWS, Google, etc.) is used in conjunction within an organization's IT solutions. In the past, consumers had questions like "Should we move to the cloud?" and "To which cloud should we move?". A shift has now occurred, where they ask "How can we take advantage of the solutions provided by multiple cloud vendors?"

As cloud adoption matures, we see core functionalities being relatively alike across major players. However, individual vendors do each offer unique services as well as vary in their ability to deliver them. For instance:

• The time to spin up a Kubernetes cluster
• The availability of low-latency networks within a certain region
• Storage costs at different performance tiers
• The compatibility of IAM services.

Multicloud solutions offer flexibility in choosing desired services. These are the 5 key concerns one should consider before opting for multicloud model.

1. Make sure the business understands the potential cost-saving aspects.
2. Obtain the necessary tooling needed to maintain clarity and reliability.
3. Start exploring multiple cloud environments to explore similarities and differences.
4. Always evaluate areas that have the potential for significant savings.
5. Continually work on automating your increasingly complex environment.

Let’s look at some examples to clarify these differences.

Serverless

With current cost considerations, Simform¹ highlights serverless costs across the three major cloud service providers (AWS, Azure, & GCP). Users are charged for using serverless functions based on the time taken to execute the code. In this hypothetical scenario, we have a serverless function that executes in a second and has 0.5 GB of RAM available. It runs 3 million times per month.

• With AWS’ serverless compute service called ‘AWS Lambda’, the total monthly costs come to around $18.74.
• The same function on Microsoft’s serverless compute service, called ‘Azure Function’, sits slightly lower at $18.
• Google’s ‘Cloud Functions’ costs about $25.15 dollars, which is about 40% more expensive when compared to the other two.

With extreme volatility in cloud computing costs, this number is bound to fluctuate sooner than later, but it does highlight the potential for incredible cost savings possible with the utilization of multicloud strategies.

Kubernetes Clusters

Another variable for differentiation is the time needed to spin up certain services. A few minutes can make a significant difference, be it troubleshooting or running iterative POCs. Let us consider the same three service providers,

• You can spin up an Amazon Elastic Kubernetes Service cluster in approximately 9 minutes.
• An Azure Kubernetes Service cluster takes about 10 minutes to complete the process.
• You can launch a Google Kubernetes Engine Cluster in 5 minutes.

In the above case, ironically, Google Cloud is about twice as fast.

So what does it look like to run a multicloud strategy? Let us look at a high-level overview of multicloud system architecture shown in the sample diagram below. Later we will dive deeper to understand what such a design could mean to an organization.

As per the diagram, you can extend your on-premise AD to the Azure AD. With that, you can authenticate and authorize the Kubernetes cluster running in the Google cloud. On the backend side, you can forward the logs from GKE and Azure to Splunk. Similarly, you can also forward the on-premise workloads to Splunk Enterprise in an AWS environment.

Cloud architecture is an ever-evolving process. The migration of associated services and solutions, also adapts to meet the changing market needs. Implementation of multicloud solutions can be daunting. In the long run, it can help organizations save significantly. You can easily re-deploy to alternate locations, provided you have flexible tooling and knowledge of diverse environments.

Potential Gains of Implementation

Data Privacy and Protection

Legislations for privacy and data protection, like the GDPR legislation in the EU, could pose challenges for organizations that are still migrating.

Going back to the example from the sample diagram discussed in the previous bit,

• Certain workloads containing sensitive data might have to be stored within the Azure compute environment in a European datacenter.
• However, using Azure AD simplifies IAM processes that run in Google Cloud and may be operating in the Kubernetes environment in North America.
• Multicloud solutions allow us to tailor data management to the needs of multiple regions and allow for compliance standards to be met.
• With Multicloud solutions, we can migrate individual workloads at a pace that matches our ability, comfort, and targets.
• It allows for increased diversification in providing their services on a global scale to easily tailor solutions for varying markets.

Reduced Vendor Lock-In Risk

• Many stakeholders exclusively use services from a single cloud provider.
• Once integrated within the provider’s ecosystem, it becomes difficult and costly to exit the ecosystem or effectively integrate it with other providers.
• Multicloud implementations can reduce the lock-in effect and increase flexibility.
• It allows workloads to be migrated across environments when necessary.
• It may be challenging if the APIs used for cross-cloud communication are not streamlined. You can run POCs for your custom architectures to ensure solution compatibility.

Global Access

• Availability of certain cloud services often comes down to the geographical location of the provider’s infrastructure.
• Despite services being similar, data centers and networking accessibility may impact key necessities such as edge computing or latency.
• Larger organizations require global access for better client experience and forced regional access may hinder it.
• Multicloud solutions can help scale applications to run within different public clouds depending on their region and availability.
• Workloads of streaming services significantly benefit from the investigation of regional data centers to improve customer experience.

Challenges and Their Mitigations

• Mutlicloud solutions, though beneficial, can be complex in implementation.
• You need manpower and in-house knowledge to take advantage of available features.
• There is a constant need to track several rapidly evolving environments.
• With changing environments, the supporting architectural decisions require constant adjustments. Hence, you may need software with a single codebase that deploys infrastructure across multiple cloud environments.
• Managing hardware and services across several clouds may lead to complex cost management. There is a constant need to stay on top of both security and reliability practices.
• More environments need increased testing, policy and identity management, redundancy etc.
• SREs and DevOps professionals need to constantly manage and observe services to keep incident response times appropriate.

In the near future, we will see numerous organizations opting for multi-cloud strategies. Hence, incorporating it in the early stages will help many organizations to edge ahead in the cloud-native future that awaits. We hope you have enjoyed reading this blog on multi-cloud environments.

‍

Squadcast is an incident management tool that’s purpose-built for SRE. Get rid of unwanted alerts, receive relevant notifications and integrate with popular ChatOps tools. Work in collaboration using virtual incident war rooms and use automation to eliminate toil.

Reference

1. https://www.simform.com/blog/compute-pricing-comparison-aws-azure-googlecloud/