Power consumption and CO2 emission of HTTP servers

Background

As global internet traffic has increased, most of the world’s internet protocol (IP) traffic goes through data centers. Greater connectivity is therefore propelling demand for data center services and energy use (mostly electricity), with multiplying effects: for every bit of data that travels the network from data centers to end users, another five bits of data are transmitted within and among data centers [3].

Companies and industries are increasingly setting voluntary efficiency and CO₂ emissions targets. In February 2020, the ICT industry agreed on a science-based target to reduce GHG emissions by 45% between 2020 and 2030. In the European Union and the United States, companies have adopted voluntary agreements to improve the efficiency of connected set-top boxes.

HTTP caches are used primarily when content is static, but are also used to cache dynamic content over short time spans. Servers that are not caches, but contain or produce the actual content, are called origin servers. The caching servers' role is to reduce the amount of traffic hitting the origin servers. For example, the main page of your local newspaper is typically cached for a short time, before the cache fetches a fresh version from the origin server. Similarly, when serving live video, video segments can be cached for a long time while other assets like video playlists can be cached for less than a second, but still drastically reduce the number of requests. Caches are generally faster and more efficient than origin servers, and can be distributed across the globe, allowing clients to fetch content from a cache near them instead of the central origin servers. A cache implementation and deployment is successful if it reduces the total cost of delivering data over HTTP and/or solves problems which are difficult to solve in the origin servers. For example, low latency is an added benefit besides cost, when using a cache.

Deliverables/Outputs

• Analyzing power consumption where content is served with and without a caching layer in front
• Developing a model for calculating the impacts of innovative ICT technologies on the power consumption and CO₂ emission of origin servers and caches
• Validating the model with caching technology developed by Varnish Software AB

This project will be carried out in collaboration with KTH, Energy Technology, division of Energy Systems, and Varnish Software AB.

Skills needed

• Experience of working in Linux based environment is needed
• Knowledge of python is required
• Please check the thesis: kth.diva-portal.org/smash/get/diva2:1697934/FULLTEXT01.pdf

Location

Varnish Software AB office, Stockholm. This position is paid.

Research Areas:

• Energy systems and innovation

Duration

6 months, start time: January 2023

How to apply

Send an email to the supervisors attaching your CV with relevant details for the position expressing your interest on the topic. Interviews will be conducted for the thesis position. This is a paid position.

KTH Supervision

Varnish Software Supervision

Clara Becker, Service Delivery Manager, clara.becker@varnish-software.com

Pål Hermunn Johansen, Product Owner of Varnish Enterprise, hermunn@varnish-software.com

Key references

1. Sandvine. The Global Internet Phenomena Report. 2020. doi: Phenomena20Report2020200507.pdf .
2. ITU (International Telecommunications Union). ICT Statistics. 2020. doi: www.itu.int/en/ITU-D/Statistics/Pages/stat/default.aspx .
3. Cisco. Cisco Global Cloud Index: Forecast and Methodology, 2016-2021. 2018. doi: white-paper-c11-738085.pdf .