Welcome to IE-LAB!

Generic filters
Generic filters

Introduction to Wavefront

Wavefront, a technology that VMware acquired in 2017, was integrated into one of VMware’s cloud services and was named Wavefront by VMware. Wavefront provides monitoring and analysis in the form of SaaS. It is designed to monitor cloud services. It can also monitor legacy systems and applications in private clouds. Wavefront collects detailed performance data and logs from various cloud services for users to Based on this, analyze the performance bottleneck of the application and quickly eliminate the application failure.

Wavefront can be used by various roles such as the platform administrator SRE (Site Reliability Engineer), developers, and system administrators. Users use the Wavefront Query Language to mine and analyze the data collected by Wavefront. The results of the analysis are presented in graphical results, which is convenient for users to analyze and compare.

Compared to traditional monitoring platforms, Wavefront has the following features:

  • Support applications of all sizes: Cloud services and distributed applications are unpredictable, and Wavefront can effectively monitor cloud services and applications of all sizes and support their flexible extensions.
  • High-speed data sampling: Containers that provide cloud services are created when needed, and are deleted immediately after use. Traditional timing sampling modes cannot meet the monitoring needs of this new generation of applications. Wavefront is designed for cloud native applications, easily supporting more than one million samples per second.
  • Self-service metrics analysis: The traditional monitoring platform is only for administrators, and Wavefront is open to all employees. Everyone can analyze the collected data using the Wavefront platform. This collaborative working mode can be better. Identify problems and motivate employees to innovate.

The Wavefront platform collects data in several ways:

  • Wavefront Agent: Wavefront has developed an agent for collecting data for various cloud services and application platforms. The Agent then uploads data to the Wavefront database in the cloud via Wavefront Proxy.
  • The monitored object sends data directly to Wavefront: you only need to insert the Metrics Library provided by Wavefront into the monitored object to collect data directly.
  • Acquire data directly from cloud services: For large cloud service platforms like AWS, Google Cloud, and Pivotal Cound Foundry, Wavefront can collect data directly.
  • Collect logs: Wavefront can also collect data from logs via TCP or FileBeat logging tools.

Wavefront currently supports a variety of mainstream cloud services and application platforms, and the supported platforms are growing rapidly, with more than 45 new integrations added in the first quarter of 2018, which you can see on Wavefront’s online documentation. New integration. So don’t worry if you don’t find the logo you need in the picture below, maybe it will be supported soon.

Wavefront application example

Let’s take a look at several examples of cloud storage service provider Boxes using Wavefront for system monitoring and data analysis, so that everyone has an intuitive understanding of Wavefront’s analysis capabilities.

One of Box’s services needs to access the object store on the public cloud. The different colored lines in the figure below represent multiple servers that provide the same service. Box uses the mechanism of anycast to achieve load balancing, but the performance of the entire service is very poor. Box found out the cause of the problem by analyzing the performance data collected by Wavefront: the network service used to implement anycast is not well balanced. On the left side of the figure we can see that these lines have large differences, and the servers represented by the purple lines are not fully utilized, which means that anycast does not play a role in balancing the load. After the problem was found, the Box was immediately fixed, and the corrected performance results showed that the performance curves of several servers tend to coincide and coincide, indicating that the load balancing mechanism is working. This type of problem can take weeks to find without a proper tool; but with Wavefront, Box’s engineers positioned the problem in a single day.

Wavefront quickly finds the problem with data analysis and gets twice the result with half the effort.

Next is an example of an operating system patching. Box’s security specification requires timely security patches for the operating system. So once new security vulnerabilities are discovered, the operations team will promptly put security patches on tens of thousands of servers. When a team member was assigned to this project, Wavefront created this metric (Metrics) to record the total number of patches that need to be applied. At the beginning, there were about 4 to 50,000 patches in total, but as the number of patches increased, the work of the project team gradually lags behind, and the accumulated unpatched ones are increasing, eventually reaching 300,000. It was impossible to get the job done on time by manpower, so they found a way to automatically patch. From April onwards, the cumulative number of patches began to plummet until July. This chart records historical real-world data and shows the team’s work performance to the management team in a very intuitive way.

Wavefront speaks with data and is truly convincing.

The following example comes from a cross-department global performance optimization project in Box. A senior leader asked the product manager to provide a report on the project decision process. The product manager then grabbed such a chart from Wavefront. The report was shared at the executive meeting attended by the CEO. What information does this chart reveal? This chart shows the performance of the Box service, with three lines representing three performance results:

  • Blue is the standard Box service, and the approximate service delay is around 1.5;
  • Green is the result of an iteration, but performance is slightly worse, and the delay rises to around 1.8;
  • Purple is the latest optimization result, reducing the service delay from 1.5 to around 0.5, which is remarkable;

Anyone can use Wavefront to mine valuable analysis results, which is the charm of Wavefront.

For more articles you can follow us on:

error: Content is protected !!
× How can I help you?