Two business women looking at a tablet computer, discussing a topic.

Embarking on the Multi-Cloud Journey

by: VMware Sr. Director of Network Services, IT Solutions Engineering and Design Swapnil Hendre, VMware IT Director Sarita Kar, and VMware Business IT Technical Architect Venkata SubbaRao

What does it mean when we say multi-cloud? When an organization uses multiple cloud computing and storage services to improve cloud infrastructure capabilities and cost, is simply known as multi-cloud.

Reason to opt for VMware Cloud on AWS

VMware IT was tasked with migrating mission-critical workloads to VMware Cloud on AWS. The objectives included extending our on-premises data center to the public cloud (hybrid cloud), achieving bidirectional workload portability, providing elasticity to scale-up or scale-down compute resources on the fly for seasonal workloads, and improving workload resiliency/availability through a consistent operational model.

Business reasons for moving to multi-cloud

We embarked on our multi-cloud journey for two main reasons—high availability and scalability. 

On the availability front, we want our services to be highly available—they should meet four 9s availability, which turns out to be a maximum of four minutes of acceptable downtime per month. To meet such uptime requirements, we expanded the application footprint to the public cloud in addition to existing on-premises data centers for redundancy reasons. 

It’s a kind of hybrid cloud that expands into multiple clouds on an as-needed basis. Hosting services across on-premises and public cloud environments have their own challenges:

  • Portability—applications need consistent environments across clouds. In our case, consistency is achieved via a cloud-agnostic toolset. We hosted our applications on VMware Tanzu® Kubernetes Grid and used VMware Tanzu® Observability by Wavefront for observability; for load balancing, we use VMware NSX® Advanced Load Balancer across on-premises and clouds.
  • A streamlined DevOps process through a unified toolset across the cloud and on-premises.
  • Data consistency, as traffic is distributed across multiple regions. We employed multiple techniques to address data consistency, pinning regional customer traffic to a geographic location, and incorporating eventual consistency patterns.

Scalability is another important aspect. For example, consider the VMware Explore event portal (formerly VMworld), which is heavily used during the three-day annual event for customers, partners, and the developer community. It needs to scale quickly for burst traffic.

To scale, we hosted the portal on VMware Cloud on AWS, which provides a consistent environment similar to an on-premises data center—no application refactoring is required. It provides good portability and a unified toolset.

The challenges of moving mission-critical services to VMware Cloud on AWS

When we decided to migrate workloads to the public cloud, we came across several challenges, such as how to establish hybrid connectivity, how to provide access to core services, how to secure workload deployed in the public cloud, how to perform data replication, and how to automate failover of traffic within minutes.

VMware IT addressed these challenges by establishing hybrid connectivity using S2S VPN first, followed by replacing it with Direct Connect. IT deployed core services locally in the public cloud to provide low latency connectivity to mission-critical applications and eliminate dependency on on-premises and egress costs. IT also performed data replication by using VMware implemented using  VMware NSX® micro-segmentation and automated failover and failback using homegrown automation.

Network and bandwidth challenges

Before committing to a full-scale migration, VMware IT made sure all the basics were in place. Initially, we used a site-to-site (S2S) VPN tunnel to quickly establish connectivity from on-premises to VMware Cloud on AWS. While that helped initially, realized that S2S VPN did not meet our business requirements of providing dedicated, low latency, cost-effective and secure connectivity. We addressed this challenge by building and employing AWS Direct Connect. 

It’s important to understand that in any implementation, you may not get it right the first time, but it is through iterations and applying different solutions that you can figure out what works best for your specific needs.

Day 2 operations

VMware IT launched several Day 2 operations initiatives to ensure the ecosystem’s ongoing viability. VMware Cloud-managed services enabled cloud administrators to focus on workload and application creation rather than infrastructure maintenance, which significantly minimized errors and reduced complexity. VMware vRealize® Network Insight monitoring delivered analytics capabilities for software-defined networking and security. This helped optimize network performance and availability with converged visibility across virtual and physical networks, along with public and private clouds. vRealize Network Insight also helped with planning application migration to VMware Cloud On AWS.

Moving to multi-cloud

Having mission-critical applications already running on VMware Cloud on AWS provides the foundation for our next goal, which is building a cloud-agnostic solution that will allow us to extend our applications to a multi-cloud environment. We are planning to deploy this platform, which will be built based on DevSecOps principles to scale and contract the workload on demand in a multi-cloud environment using various VMware products such as Tanzu Kubernetes Grid, VMware NSX® Advanced Load Balancer, VMware SD-WAN, vRealize Network Insight, vRealize Log Insight, and vRealize® Automation

Business benefits of moving to multi-cloud

  • Uber goal is to provide customers with unhampered & consistent services, any downtime of our services will affect customers’ business SLAs. 
  • The Multi-cloud journey helped systems to react quickly to customer usage patterns, as systems need to autoscale, without preplanning.
  • Another benefit is fulfilling regulatory needs & risk eversion and serving content from locations near customers’ geographic locations, which helps to reduce latency, improved customer experience, fulfilling data localization needs.

Cost-benefit analysis from a business perspective

Yes, hyperscalers provide auto-scalability on-demand basis. Our service usage usually has sudden spikes based on campaigns & events. That means, we have peak traffic for some stipulated time and becomes flat. Cloud provides elasticity; we can ram up resources and ramp down quickly, without upfront investment. 

Also adopting a cloud-agnostic toolset, we can always switch to cost-effective clouds anytime and not be permanently hooked to one cloud provider.

For more information, listen to the fourth episode of our Digital Transformation series.

VMware on VMware blogs are written by IT subject matter experts sharing stories about our digital transformation using VMware products and services in a global production environment. Contact your sales rep or to schedule a briefing on this topic. Visit the VMware on VMware microsite, read our blogs and IT Annual Report and follow us on Twitter.


Leave a Reply

Your email address will not be published. Required fields are marked *