In our blog “Why Should You Go Serverless in the Cloud,” we explain how serverless computing isn’t really serverless at all. The event-driven application design and development programming functions of a serverless model still run on server and storage hardware. The term “serverless” actually speaks to how DevOps teams no longer have to worry about provisioning server compute and storage resources for their application development and deployment because they can now access everything they need as a scalable cloud service.
Serverless cloud computing is offered by all the major cloud providers (AWS, Microsoft Azure, Google, Oracle, IBM, etc.) and they provide developers with useful APIs to control a cloud platform’s development infrastructure and its components with tremendous flexibility. You can run an application’s code on the same cloud hardware resources. Or split your application development into discrete functions, including running a task in response to events, across multiple services on the same cloud platform or across multiple cloud providers. This would allow you to run application components using the best cloud service for each development activity.
An automated public or private cloud-based resource scheduler manages how, when and where to run and scale various application components in a “provision-as-you go” model. You can also leverage containers to orchestrate the development of different application components across cloud services. So the question for developers becomes, how do you best tap into these serverless cloud resources and services for the fastest and most reliable application development and deployment experience?
Building the best infrastructure for serverless computing
Serverless computing in the cloud is a distributed infrastructure, with whole applications or pieces of applications, running on multiple clouds. According to Joshua McKenty, vice president of systems advisory group at Pivotal, “As a developer, I never want to know too much about infrastructure, but if you’re building a distributed system, some of your code probably runs on a different server and there’s a network hop in there somewhere. If you’re writing serverless, or let’s call it functions as a service, your functions are probably in different servers. So understanding the network, at least, is important.”[i]
The “hop” that McKenty is speaking of is when application traffic passes through a string of routers as it travels along a wide area network (WAN). In fact, many times, your cloud service provider is routing your traffic to a cloud that is in a completely different city from where your data center resides. This could mean multiple hops along the way, with increases in latency (delays) that slows communications with the cloud and makes services less responsive. This could create a very frustrating situation in a DevOps environment, especially if you are trying to coordinate various application development functions across multiple clouds.
Many of our customers solve this dilemma by deploying a distributed hybrid multicloud infrastructure using vendor-neutral interconnection hubs, such as the Equinix Cloud Exchange Fabric (ECX Fabric), in metros with the greatest number of their chosen cloud service providers. This enables them to directly and securely interconnect with any cloud service provider on Platform Equinix® within our more than global 200 data centers.
This environment is perfect for those multinational enterprises with application developers scattered all over the world that also need to interact and collaborate with each other and quickly access cloud resources. In most cases, if you are in the same data center facility as your cloud provider on Platform Equinix, their services are less than 10 milliseconds away. And in those situations where the cloud provider of choice is not in the same metro as your development or deployment environment, then you can interconnect ECX Fabric across multiple Equinix data centers to gain access to the clouds you require.
Ensuring serverless computing availability
In a rare instance, your serverless computing cloud service may not be available. Though most public cloud infrastructures are more robust than many enterprises’ data centers, there are service level agreements (SLAs) attached to serverless computing in the cloud that differ from what you may be getting when purchasing Infrastructure-as-a-Service (IaaS). Though downtime happens less frequently in serverless computing platforms than it does with IaaS, serverless SLA contracts typically allow for several hours of uncompensated service interruptions each month.[ii] Some cloud providers offer credits to make up for the downtime, but it would be better to proactively include serverless computing in your cloud high availability (HA) strategy to limit any downtime.
You can quickly and easily provision connections between the same cloud provider’s services in two separate metros for HA or between two different cloud platforms in the same metro to ensure continuous serverless computing access.
Provisioning network connections between two or more clouds using ECX Fabric software-defined interconnection can be done in minutes. You can also harness virtual network services using Network Edge and select, configure and connect virtual network devices from industry-leading vendors like Cisco, Juniper and Palo Alto Networks without requiring additional space, power, hardware or equipment. Once this infrastructure is in place, you have a reliable, high availability environment for serverless application development and testing in the cloud. The same environment will also protect your application availability once you go into production.
Serverless computing in the cloud provides infinite flexibility and scalability for DevOps organizations. And the increased performance and reliability of a hybrid multicloud infrastructure, with direct and secure interconnection, will enable you to accelerate your new application and solution time to market exponentially.
Learn more about ECX Fabric and how you can create hybrid multicloud infrastructures.
You may also want to check out our Cloud Optimization Workshop.
[i] TechTarget SearchCloudComputing, “In a serverless architecture age, infrastructure still matters,” May 2018.
[ii] TechTarget SearchCloudComputing, “Build an event-driven cloud app with serverless SLAs in mind,” April 2019.