Monitoring Multi-Cloud Network Performance

Posted by on June 21st, 2018
June 21st, 2018

Most enterprises began their cloud journey with a single infrastructure-as-a-service platform. Lately, compelling offerings from the likes of Amazon Web Services (AWS), Microsoft Azure, Google Cloud (GCP), IBM, Digital Ocean and others have created a plethora of options, with best-of-breed functionality designed for different workload types. Much as we saw with computing and server platforms in private data centers, enterprises are finding themselves consuming two or more of these cloud services. More recently, we are seeing enterprises embrace multi-cloud right out of the gate as a primary cloud-first strategy. Many factors dictate this. For some, it’s a strategic decision to help avoid vendor lock-in, manage costs and access best-of-breed functionality. For others, it’s an evolutionary journey dictated by changing workload needs. For still others, it’s the promise of a seamless containerized cloud using technologies like Kubernetes. However one thing is very clear from our conversations with customers—adoption is NOW. In a recent survey we conducted at the Cisco Live conference in Orlando, over 32% of respondents reported a strategic multi-cloud initiative at their companies, and 28% of the respondents already identified as multi-cloud users.

This proliferation of IaaS providers and modular application architectures have resulted in a complex matrix of inter-service communication across infrastructures and networks that enterprises do not own or control. Much of this communication traverses the Internet, which has evolved into a mission-critical transport for enterprises. Complexity has the potential to drive up cost if not managed carefully. Comprehensive visibility and monitoring are critical to ensuring a good digital experience for customers and employees.

ThousandEyes Network Intelligence coverage has always included a vast array of global vantage points both inside and outside the enterprise environment. We have now expanded this coverage to include Cloud Agents in 15 AWS regions, 15 Google Cloud regions and 25 Microsoft Azure regions. In addition, we have now enabled the ability to configure Agent-to-Agent tests between Cloud Agents. Previously, Agent-to-Agent tests required an Enterprise Agent on at least one end of the test. With these enhancements, organizations leveraging any combination of Amazon Web Services (AWS), Microsoft Azure and Google Cloud Platform (GCP), have the ability to measure and visualize app and network-layer performance metrics on a cloud-to-cloud, Internet-to-cloud and inter-region basis.

Simplifying Multi-Cloud Visibility

One of the first decisions facing a cloud deployment is where to locate your workloads. Most major IaaS providers have a multitude of data center locations across the globe. It is important to understand network latencies and external dependencies before deploying any production resources or even signing contracts with these providers. With ThousandEyes Cloud Agents, one can easily setup bi-directional Agent-to-Agent Network Tests between 150+ cities across the globe, and 55 IaaS data centers to determine the best fit.

Latency chart for multiple Cloud providers
Figure 1: Latency comparison from multiple cloud providers to London, UK.

In the example above, we are looking at a time series Dashboard widget showing latency from four cloud data centers to London, UK. This enterprise has a significant user base in London, UK hence any workload decisions must factor in application performance for these users. It’s very clear from this Dashboard that the AWS us-east-1 and GCP us-east-4 are consistently better performers from a latency perspective.

Another use case may involve the consumption of external API services which happen to be hosted on an IaaS platform. Two great examples of these are Azure AD and Amazon’s S3 service, hosted on Azure and AWS respectively. These are widely consumed by many enterprises and SaaS applications.

HTTP Server view for Amazon S3
Figure 2: Application layer view into Amazon’s S3 service.
Path Visualization with AWS, Azure and GCP
Figure 3: Service delivery paths from AWS, Azure and GCP to Amazon S3.

The example above shows application availability, network paths and latencies from several cloud regions to Amazon’s S3 service in the us-east-1 region. This is an important visibility dimension for anyone building or consuming cloud-native applications that use the S3 service. It’s not only important to understand and trend performance, but also understand root cause when the service fails.

Enterprise Agents for Enhanced Multi-Cloud Visibility

While Cloud Agents greatly simplify the ability to visualize multi-cloud dependencies, customers have always been able to deploy ThousandEyes Enterprise Agents on any of the IaaS platforms. Our Enterprise Agents can be deployed as virtual machines, Linux packages or Docker containers on virtually any cloud service with a few simple steps. This gives customers a view into their own VPCs and private networks. Enterprise Agents have a very small footprint and do not require significant compute resources. Unless you’re running complex Transaction Scripts, in most cases you may be fine with as little as one core and 2GB RAM (please refer to our knowledge base for the most up-to-date guidance).

Enterprise Agent Linux Package install instructions
Figure 4: Easily deploy ThousandEyes Enterprise Agents on any cloud service.
Enterprise Agents Docker instructions
Figure 5: Deploy ThousandEyes Enterprise Agents in Docker containers using pre-defined templates.

One of the simplest and quickest ways to deploy Enterprise Agents in a cloud platform is to spin up an Ubuntu 16.04LTS virtual machine and ensure it is connected to a virtual network with outbound Internet access. With three simple CLI commands, you can have an Enterprise Agent running on your cloud in a matter of minutes. For larger scale deployments, the whole process can be automated using server automation tools like Chef or Puppet. Here is a handy cookbook for deploying Enterprise Agents using Chef. In the next few sprints, we will be making this process even simpler by leveraging pre-populated templates that will allow users to launch Enterprise Agents on GCP, AWS and Azure directly from within the ThousandEyes UI.

Multi-cloud path visualization
Figure 6: Path Visualization showing traffic paths from Cloud Agents to an Enterprise Agent in Google Cloud.

The example above shows network paths from various cloud data centers to an Enterprise Agent running on Google Cloud.

Together with our pre-deployed Cloud Agents, ThousandEyes’ Network Intelligence provides enterprises with visibility into all cloud infrastructure across any network, thus overcoming the complex operational challenges of multi-cloud deployments to accelerate cloud adoption, streamline IT operations and deliver uninterrupted and superior digital experiences. Request a demo today to see how ThousandEyes can help your organization embrace a successful multi-cloud strategy.

Processing...