The Lockdown Effect: Implications of the COVID-19 Pandemic on Internet Traffic Oliver Gasser, Max Planck Institute for Informatics Š Virtual Š 27-30 Oct 2020 RIPE81 Meeting
COVID-19 and the Internet
COVID-19 and the Internet 1
COVID-19 and the Internet The Internet is essential in all these efforts, but how well does it cope? 1
Goals of this presentation • Understand the impact of the COVID-19 pandemic on different types of networks 2
Goals of this presentation • Understand the impact of the COVID-19 pandemic on different types of networks • Present our results from different networks in the core and at the edge 2
Goals of this presentation • Understand the impact of the COVID-19 pandemic on different types of networks • Present our results from different networks in the core and at the edge • Get to know your experience during COVID-19 2
Lots of data • Edge network: Large European ISP • Core networks: 3 IXPs in Central Europe, Southern Europe, and at US East Coast • Academic network: REDIMadrid university network in Madrid
Lots of data, lots of data crunchers • Edge network: Large European ISP • Core networks: 3 IXPs in Central Europe, Southern Europe, and at US East Coast • Academic network: REDIMadrid university network in Madrid Anja Feldmann Oliver Gasser Franziska Lichtblau Enric Pujol Ingmar Poese Christoph Dietzel MPII MPII MPII BENOCS BENOCS DE-CIX Daniel Wagner Matthias Wichtlhuber Juan Tapiador Narseo Vallina Oliver Hohlfeld Georgios Rodriguez Brandenburg University Smaragdakis DE-CIX DE-CIX Universidad Carlos III de of Technology Madrid IMDEA, ICSI TU Berlin, MPII
Traffic changes in different networks
Traffic changes from January 2019 to June 2020 1.6 ISP, Europe (>15M fixed-network lines) COVID-19 outbreak in Europe Initial responses and lockdowns 1.4 Normalized traffic volume 1.2 1.0 0.8 0.6 2019 2020 0 2 4 6 8 0 2 4 6 8 0 2 4 6 8 0 2 4 6 8 0 2 4 6 8 0 2 2 4 6 8 0 2 4 6 8 0 2 4 6 8 1 1 1 1 1 2 2 2 2 2 3 3 3 3 3 4 4 4 4 4 5 5 1 1 1 1 1 2 2 2 2 2 Calendar week 4
Traffic changes from January 2019 to June 2020 1.6 ISP, Europe (>15M fixed-network lines) COVID-19 outbreak in Europe Initial responses and lockdowns 1.4 Normalized traffic volume 1.2 1.0 0.8 0.6 2019 2020 0 2 4 6 8 0 2 4 6 8 0 2 4 6 8 0 2 4 6 8 0 2 4 6 8 0 2 2 4 6 8 0 2 4 6 8 0 2 4 6 8 1 1 1 1 1 2 2 2 2 2 3 3 3 3 3 4 4 4 4 4 5 5 1 1 1 1 1 2 2 2 2 2 Calendar week Once the lockdown started the ISP saw a +30% increase in traffic which normally spans over multiple months. 4
Traffic changes from January 2019 to June 2020 1.6 ISP, Europe (>15M fixed-network lines) COVID-19 outbreak in Europe IXP, Central Europe (>900 members) IXP, South Europe (>170 members) Initial responses and lockdowns 1.4 IXP, US East Coast (>250 members) Normalized traffic volume 1.2 1.0 0.8 0.6 2019 2020 0 2 4 6 8 0 2 4 6 8 0 2 4 6 8 0 2 4 6 8 0 2 4 6 8 0 2 2 4 6 8 0 2 4 6 8 0 2 4 6 8 1 1 1 1 1 2 2 2 2 2 3 3 3 3 3 4 4 4 4 4 5 5 1 1 1 1 1 2 2 2 2 2 Calendar week Similar behavior for the IXPs; for the IXP CE and IXP US the traffic levels remain elevated. 4
Traffic changes from January 2019 to June 2020 1.6 ISP, Europe (>15M fixed-network lines) COVID-19 outbreak in Europe IXP, Central Europe (>900 members) IXP, South Europe (>170 members) Initial responses and lockdowns 1.4 IXP, US East Coast (>250 members) Mobile operator, Europe (>40M customers) Normalized traffic volume 1.2 1.0 0.8 0.6 2019 2020 0 2 4 6 8 0 2 4 6 8 0 2 4 6 8 0 2 4 6 8 0 2 4 6 8 0 2 2 4 6 8 0 2 4 6 8 0 2 4 6 8 1 1 1 1 1 2 2 2 2 2 3 3 3 3 3 4 4 4 4 4 5 5 1 1 1 1 1 2 2 2 2 2 Calendar week Once the lockdown started mobile traffic decreased measurably and increased again with the first relaxations in mid April. 4
Changes in workday vs. weekend patterns at the ISP Wednesday Feb 19 1.0 Normalized traffic volume 0.8 0.6 0.4 0.2 0.0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Hour of the day 5
Changes in workday vs. weekend patterns at the ISP Wednesday Feb 19 1.0 Normalized traffic volume 0.8 0.6 0.4 0.2 0.0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Hour of the day • Regular patterns • Workday: Strong increase in evening hours 5
Changes in workday vs. weekend patterns at the ISP Wednesday Feb 19 1.0 Saturday Feb 22 Normalized traffic volume 0.8 0.6 0.4 0.2 0.0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Hour of the day • Regular patterns • Workday: Strong increase in evening hours • Weekend: More traffic during daytime 5
Changes in workday vs. weekend patterns at the ISP Wednesday Feb 19 1.0 Saturday Feb 22 Normalized traffic volume Wednesday Mar 25 (lockdown) 0.8 0.6 0.4 0.2 0.0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Hour of the day • Regular patterns • Workday: Strong increase in evening hours • Weekend: More traffic during daytime 5 • During lockdown: Workdays look more like weekends
Changes in workday vs. weekend patterns at the ISP correct weekend wrong correct workday wrong • Classify days into workdays or weekends using traffic patterns 6
Changes in workday vs. weekend patterns at the ISP correct weekend wrong correct workday wrong • Classify days into workdays or weekends using traffic patterns • Pre-lockdown: Most days are classified correctly 6
Changes in workday vs. weekend patterns at the ISP correct weekend wrong correct workday wrong • Classify days into workdays or weekends using traffic patterns • Pre-lockdown: Most days are classified correctly • During lockdown: Workdays are classified as weekends 6
Changes in workday vs. weekend patterns at the ISP correct weekend wrong correct workday wrong • Classify days into workdays or weekends using traffic patterns • Pre-lockdown: Most days are classified correctly • During lockdown: Workdays are classified as weekends; recovering after mid-May 6
Changes in workday vs. weekend patterns: ISP vs. IXP ISP IXP } } Norm. traffic volume Sat-Sun Norm. traffic volume Sat-Sun with weekend-like pattern with weekend-like pattern Mon-Fri Mon-Fri 1.00 1.00 0.75 0.75 0.05 0.05 0.25 0.25 Norm. traffic volume Norm. traffic volume 0.00 0.00 0.00 0.00 0.25 0.25 0.05 0.05 0.75 0.75 } } 1.00 1.00 Mon-Fri Mon-Fri with workday-like pattern with workday-like pattern Sat-Sun Sat-Sun Jan 1 Feb 1 Mar 1 Apr 1 May 1 Jun 1 Jan 1 Feb 1 Mar 1 Apr 1 May 1 Jun 1 Date Date 7
Changes in workday vs. weekend patterns: ISP vs. IXP ISP IXP } } Norm. traffic volume Sat-Sun Norm. traffic volume Sat-Sun with weekend-like pattern with weekend-like pattern Mon-Fri Mon-Fri 1.00 1.00 0.75 0.75 0.05 0.05 0.25 0.25 Norm. traffic volume Norm. traffic volume 0.00 0.00 0.00 0.00 0.25 0.25 0.05 0.05 0.75 0.75 } } 1.00 1.00 Mon-Fri Mon-Fri with workday-like pattern with workday-like pattern Sat-Sun Sat-Sun Jan 1 Feb 1 Mar 1 Apr 1 May 1 Jun 1 Jan 1 Feb 1 Mar 1 Apr 1 May 1 Jun 1 Date Date • At both vantage points workdays are mostly classified as weekends 7
Poll Q: Did the traffic in YOUR network change during the lockdown? • Increase in traffic (more than 10%) • No significant change (0–10%) • Decrease in traffic
Application-level traffic changes
Classify traffic by application class • Classify based on transport ports and src/dst ASes 9
Classify traffic by application class • Classify based on transport ports and src/dst ASes Base week IXP Web conf Coll. working Email VoD Gaming Social media Educational CDN u n n d i t e h r a u o u e F W T S S M T 9
Classify traffic by application class • Classify based on transport ports and src/dst ASes Base week IXP Web conf Coll. working Email VoD Gaming Social media Educational CDN u n n d i t e h r a u o u e F W T S S M T • Email during working hours 9
Classify traffic by application class • Classify based on transport ports and src/dst ASes Base week IXP Web conf Coll. working Email VoD Gaming Social media Educational CDN u n n d i t e h r a u o u e F W T S S M T • Email during working hours • Video, gaming, and social media during evening hours 9
Classify traffic by application class • Classify based on transport ports and src/dst ASes Base week IXP Web conf Coll. working Email VoD Gaming Social media Educational CDN u n n d i t e h r a u o u e F W T S S M T • Email during working hours • Video, gaming, and social media during evening hours • Hardly any web conferencing 9
Changes in application classes: Central European IXP 10
Changes in application classes: Central European IXP March: • Increase in web conf., VoD, and gaming • Partial decrease in CDN and educational traffic 10
Recommend
More recommend
