On this page, we present the recorded measurement data for our large scale assessment of global HTTP/2 adoption, based on
i) scanning the entire IPv4 address space and ii) the complete .com/.net/.org domains.
In our papers, we analyzed how HTTP/2 Server Push, a feature that promises to further reduce the page load time, is utilized in the wild. In the absence of standardization and recommendations for concrete strategies, we identified current undocumented practices having potential influence on page load time and network resource usage, and how this impacts the human perception. Finally, we analyze the performance impacts of Server Push in a controlled testbed and, based on the results, propose a new server scheduler, that interleaves resources via push to improve the visual progress.
Results & Publications
HTTP2 Adoption, User Peception and Performance
We publish HTTP/2 (Server Push) adoption statistics on this website on a regular basis. The lists of discovered HTTP/2 (push domains) can also be downloaded.
Moreover, (concise) results of our study on end-user perception as well as the analysis of Server Push performance impacts and the concept of Interleaving Push are available.
Information on our methodology and further results can be found in the following publications:
- 2019-04-18: We improved the stability of our scanners. In addition, we now also show the HTTP/2-enabled infrastructures (ASNs) in our measurements.
- 2018-09-13: Our paper Is the the Web ready for HTTP/2 Server Push? got accepted at ACM CoNEXT 2018 (pre-print).
- 2018-06-06: More information and results of our joint work (TU München, Universidad Carlos III de Madrid/IMDEA, ICSI and RIPE NCC) on the significance, structure and stability of Internet top lists is available here.
- 2018-06-06: We finished the maintenance on our DNS infrastructure, but unfortunately some data points for the smaller lists (beginning of June) are missing.
- 2018-03-20: We presented our work on HTTP/2 Server Push at the IETF 101 meeting in London as part of the Measurement and Analysis for Protocols Research Group.
- 2018-02-20: We observe an increase in pushing websites, mainly hosted by Cloudflare. However, we believe that this is caused by the use of the HubSpot Content Optimization System, which generates preload headers, which in turn are pushed by Cloudflare.
- 2018-01-05: Due to a request, we excluded some IPs from our measurements, causing a drop in the overall adoption figures.
- 2017-10-31: We finished the maintenance work on our measurement infrastructure.
- 2017-10-24: Our measurement infrastructure is currently under maintenance and is back online soon. We apologize for the inconvenience.
- 2017-10-06: We observe that the domain parker (responsible for >6K pushing domain in the .com set) seems to experience SSL problems (ERR_SSL_UNRECOGNIZED_NAME_ALERT), causing a drop of push enabled websites in our measurements.
- 2017-09-22: We added the .de (incomplete zone!) set to our measurements.
- 2017-09-22: Apparently, Cloudflare reactivated Server Push for their customers, as observable in our Adoption Results.
- 2017-09-08: We experience a significant drop in websites utilizing HTTP/2 Server Push among several data sets, most of them hosted by Cloudflare. Update: Cloudflare has temporarily deactivated the Server Push feature for maintenance (Source).
- 2017-09-04: irs.gov is the first HTTP/2 Server Push capable .gov domain. irs.gov enabled HTTP/2 support on Wed, Aug 30th and started to push 2 stylesheets on Fri, Sep 1st. The site is hosted by Akamai.
- 2017-09-03: We added the .at (incomplete zone!) set to our measurements.
- 2017-05-17: We added the .gov set to our measurements.
- 2017-05-06: Our paper A QoE Perspective on HTTP/2 Server Push got accepted at ACM SIGCOMM Internet-QoE 2017.
- 2017-04-07: Our paper How HTTP/2 Pushes the Web: An Empirical Study of HTTP/2 Server Push got accepted at IFIP Networking 2017.
- 2017-01: We started scanning various domain lists (Alexa 1M, .com, .net, .org, .fi, .se, .nu) and analyze the HTTP/2 (Server Push) Adoption.
This work has been funded by the DFG as part of the CRC 1053 MAKI and the European Union’s Horizon 2020 research and innovation program 2014–2018 under grant agreement
No. 644866 (SSICLOPS). It reflects only the authors’ views and the European Commission is not responsible for any use that may be made of the information it contains.