How To Solve For…

Solve Network Limitations to Achieve Your Multicloud Potential

Learn how Uber completed a massive cloud-to-cloud data migration by choosing the right multicloud networking solution

Arun Dev
Raam Muthusamy
Solve Network Limitations to Achieve Your Multicloud Potential

TL:DR

  • Multicloud networking challenges such as diverse cloud architectures, security vulnerabilities & latency issues strain enterprise resources.
  • Uber used Equinix Fabric Cloud Router to migrate 6.5 petabytes between clouds, meeting their aggressive timeline cost-effectively.
  • On-demand, direct cloud-to-cloud connectivity accelerates data migrations, while eliminating the need for equipment purchases & long-term contracts.

Using multiple public clouds is considered mainstream among enterprises, yet many Infrastructure teams struggle to manage their multicloud strategies effectively—for valid reasons. To start with, every cloud service provider (CSP) has its own networking architecture and connectivity framework that enterprise Infrastructure teams need to learn, support and manage. Other contributing factors include challenges with connectivity, security, management complexity, performance, reliability, interoperability and integration.

Has your experience with multicloud networking been filled with complexity that’s distracting, disruptive and costly? Undoubtedly, managing cloud infrastructure and multicloud networking can be hard, but it’s possible to eliminate some of the complexity.

If you attended our session at Google Cloud Next 25, you heard a compelling story about how Uber completed a massive data migration from one public cloud to another public cloud. We’ll recap that migration in the blog post below.

Overcoming networking challenges in multicloud environments

To ensure the success of your multicloud strategy, there are several networking challenges that you may need to solve.

Diverse cloud environments strain already-stretched resources: Running multicloud environments involves more than clouds. It requires juggling legacy and modern systems across multiple clouds, knowing each cloud’s different tools and configurations, and possessing the skills to understand the differences between the clouds due to the lack of industry standards.

Limited visibility increases security risks and compliance gaps: Managing data spread across multiple clouds requires ensuring consistent policies and real-time tracking of data updates. Oversight is critical to managing risk and meeting compliance requirements.

Latency issues impact user experience: Ensuring performance is another challenge that affects all industries. When data gets transferred from one cloud to another, the slightest delay can lead to frustrated customers.

Inconsistent security frameworks introduce vulnerabilities: Each cloud manages security differently, which can create vulnerabilities. Without consistent policies across multicloud environments, enterprises face compliance risks that could lead to costly fines.

Success with networking in a distributed multicloud environment can be achieved by applying the following strategies:

  • Make sure you’re in the right locations, close to the intersection of private networks, public clouds and the internet.
  • Engage with the right partners and providers, close to cloud on-ramps with an ecosystem where you can collaborate and innovate.
  • Establish a unified global network policy to ensure consistent security and performance while simplifying management, reducing operational costs and streamlining upgrades.

Use case: Solving for a large cloud-to-cloud migration at Uber

Every multicloud networking scenario is different, whether it involves a one-time migration between clouds or the ongoing support of distributed multicloud environments. The right solutions can deliver lower latency, improved end-user experience and cost savings.

Take Uber, for instance. As the world’s largest ridesharing company, with 150 million monthly active users across 70 countries and 15,000 cities, they faced the daunting task of migrating 6.5 petabytes of diverse data from a public cloud to Google Cloud—quickly.

To put this volume of data into perspective, consider that it equals:

  • Approximately 6.5 billion photos
  • Roughly 4.6 million hours of HD video, which equates to more than 525 years of continuous video playback
  • Around 6.5 trillion pages of standard text documents

The migration requirements were stringent:

  • Completely isolate this migration from existing physical network connections.
  • Avoid the risk of compromising the performance of routine production traffic running on the backbone.
  • Prevent cloud connectivity from becoming saturated with migration traffic.

Cost was also a significant factor; Uber would incur duplicate data storage costs on two public clouds during migration, so ensuring a fast transfer was crucial.

Uber was under intense pressure to complete the migration as quickly as possible. Uber explored several options which were not ideal. They risked overwhelming the Uber backbone or paying expensive public internet egress fees. When Uber started discussions with Equinix, they needed to be up and running in under a week. Reliability was crucial, as any downtime would increase cloud costs.

Ultimately, it became clear that the Equinix Fabric Cloud Router solution was the optimal choice for our migration, as it would provide the direct cloud-to-cloud connectivity and low latency we needed without locking us into a long-term contract. Since Fabric Cloud Router is an on-demand solution, we could simply turn down the service once the migration was complete.” Raam Muthusamy, Senior Staff Infrastructure Engineer, Uber Technologies, Inc.

Equinix Fabric Cloud Router allows dynamically scaling up the bandwidth (from 20 Gbps to 40 Gbps) and scaling down as needed. Uber was able to spin up the service in less than a day. And they met their aggressive timeline for launching the migration project.

As demonstrated by Uber’s experience, multicloud networking no longer has to be complex.  Alternatively, Uber would have had to buy networking equipment or multi-year circuits from carriers to perform simple cloud migrations. Instead, they completed the migration using an as a Service solution that was easy, fast and cost-effective.

Solving multicloud networking challenges with Equinix and Google Cloud

Deploying digital infrastructure at Equinix makes it easy to connect with Google Cloud. As a leading provider of cloud on-ramps, Equinix can help enterprises access Google Cloud in 44+ metros; in fact, 20% of Equinix customers are already doing so.

Equinix and Google Cloud don’t just leverage each other’s technology: We’re working together to make the cloud journey quicker, simple and more effective for our joint customers. They can optimize the performance of applications running in Google Cloud and enhance user experiences that improve workplace productivity.

Here are just a few of the ways we connect customers with Google Cloud from Equinix locations:

  • 24+ Equinix metros have Google Cloud low-latency regions.
  • 19+ Equinix metros have cross-cloud interconnect access.
  • 32+ Equinix metros have high-speed virtual connections of 20/50 Gbps.

Equinix simplifies multicloud networking with solutions that ensure seamless integration across multiple cloud environments, optimizing performance and reducing costs.

Discover how Equinix helps drive value and streamline multicloud operations: Read the Multicloud Networking Guide.

 

You might also be interested in:

Equinix Fabric Cloud Router Datasheet

 

Avatar photo
Arun Dev Vice President, Digital Interconnection Services
Avatar photo
Raam Muthusamy Guest Author: Senior Staff Infrastructure Engineer, Uber
Subscribe to the Equinix Blog