The DeathStarBench suite is an open-source application suite developed at Cornell University. It is built with microservices that are representative of real world, end-to-end, modular, extensible services.

• The socialNetwork service is built using popular open-source applications commonly deployed by cloud application providers, such as NGINX, Memcached, MongoDB. The application is implemented with loosely coupled microservices, communicating with each other via Thrift RPCs and written in various languages including C/C++, Python and Lua scripting.

• Users (clients) send requests over http, which first reach a load balancer. Once a specific webserver is selected, NGINX service uses a Lua module to talk to the microservices responsible for composing and displaying posts. The service’s backend uses Memcached/Redis for caching, and MongoDB for persistent storage for posts, profiles, media, and recommendations.

• The DeathStarBench suite is packaged with a modified workload generator WRK2, which is based on WRK. WRK2 uses LuaJIT scripts to perform HTTP request generation, response processing, and custom reporting. The script used for testing is Compose Post which simulates clients connecting to the social network applicaiton. The suite composes posts that include text, media, and shortened URLs , which get compiled, saved and added to both home and user timelines.

The figure below shows the dependency graph for the microservices that make up the social network application

• The SocialNetwork application can be deployed on a single-socket bare metal server or cloud compute instance using Kubernetes.

• Once the social network application is installed on the test system, helm charts are used for quick and easy installation of the micro-services on the Kubernetes cluster.

• The client workload generator WRK2 is run on a seperate server/virtual-machine on the same internal network as the test system.

• The WRK2 test scripts generate http requests starting with a fixed number of threads (simulated connections) and requests per second.

• The load is gradually increased by dialing up the requests per second.

• The resulting latency and throughput are measured after every run.

WordPress represents a typical multi-tier application that uses the open-source solution stack called LAMP stack. The LAMP stack is primarily used in web development and in this solution, we deploy Wordpress on a LAMP stack using Kubernetes.

We use Wordpress as a solution to showcase the power of cloud-native applications and containerized deployments on Ampere processors. Each component in this deployment has its own container(s). The WordPress container is the frontend tier, and the MySQL container is the database/backend tier for WordPress.

• WordPress
• MySQL
• Kubernetes Cluster
• Load Tester (K6)

The goal of this solution is to benchmark the performance of a WordPress hosting website and determine the peak performance. Peak performance is measured using a load testing service like K6 which emulates large numbers of users visiting a website and watching how well the hosting responds under these stressful conditions.

All tests were performed on an identical WordPress dummy website with the same plugins and same dummy site installed on a Virtual Machine. The virtual machines used for competitive analysis are Ampere Altra, Intel IceLake and AMD Milan

The test uses a separate client virtual machine for running the load generator. The client can reach the public ip of the WordPress site that’s deployed.

• Script simulates users access the blog at http://ip-address-of-wordpress-site
• Increase the load by increasing number of users from 100 to 300, in steps of 20
• Test Run duration is 5 mins
• Each test is run at least 5 times and the run with highest P99 latency is used

At the end of each run, the K6 metrics return the following:

• P99 Latency
• HTTP requests/sec
• Failed HTTP Requests

Plesk is the leading WebOps platform to build, secure and run websites, applications and hosting businesses. A single control panel with an intuitive graphical interface, a ready-to-code environment and powerful extensions.

Key solution areas include:

• Unlimited Domains & DNS integrated
• WordPress Toolkit with automated staging/cloning included
• Laravel Toolkit, .NET on Linux Toolkit, improved GIT integration
• Support for NodeJS, Ruby, Docker and LAMP stacks w/ Apache or NGINX
• Security & backups across all levels of a website or application stack
• Subscription Management, Account Management and Reseller Management
• 100+ Plesk Extensions in the in-app catalog

Deploying Plesk on OCI Ampere A1 Instances - Video Tutorial

• "From Initial exploration to public beta, our engineering team was able to create a POC running on OCI Ampere A1 within six weeks."
  • Lukas Hertig, Sr. Vice President of Business Dev. and Strategic Alliances at Plesk/WebPros See Case Study
• Oracle Cloud Marketplace: Plesk
• Visit Plesk.com to learn more