RIPE Database Working Group Minutes RIPE 82

Session: Thursday, 20 May, 16:00 – 17:00
Chairs: Denis Walker, William Sylvester
Scribe: Boris Duval
Status: Final

A. Introduction

William Sylvester welcomed attendees and thanked the RIPE NCC for their support.

B. Operational Update RIPE Database

Edward Shryane, RIPE NCC

Presentation available at:
https://ripe82.ripe.net/archives/video/607/

Cynthia Revström shared that she was supporting Ed’s initiative to email the top 10 LIRs with the most PERSON objects and that she was keen to hear their feedback.

Wessel Sandkuijl (Prefix Broker BV) referred to a presentation from Remco van Mook who pointed out that there were objects with typos in the status attribute in the RIPE Database. Wessel asked if there were plans to fix this issue.  

Ed said that the database is case-insensitive and that this explains the different typos found by Remco. He added that if this was an issue for the community, the RIPE Database Team will be looking into it and conduct an impact analysis.

Randy Bush asked to talk with the RIPE NCC legal department about geofeed PII implications.

Ed mentioned that the RIPE NCC legal department were already aware of this issue and that he will put Randy in touch with them.

Job Snijders asked if RIRs were using a detached signature to verify they retrieved the delegated stats from the RIR. He said that he saw that a simple MD5 checksum exists but added that a GPG signature on the MD5 checksum could improve assurances when delegated stats are to be used to remove RIPE-NONAUTH objects.

Ed mentioned that the RIPE NCC is currently using the MD5 checksum and that you can connect over to HTTPS to retrieve delegate stats. However, he added that it’s always a good idea to check using a GPG signature.

Daniel Karrenberg (RIPE NCC) asked if stopping the 'white pages' meant that DK58 will be deleted.

Ed said that the issue is whether the community want to maintain the list or if they want to delegate it to the RIPE NCC.

Cynthia commented that she would prefer the exclusion list over white pages. But regardless of what is used, she added that the community should choose between having white pages or an exclusion list.  

Ed agreed with Cynthia’s comment.

Christian Bretterhofer (Independent) commented that he saw a lot of inetnum and inetnum6 object referencing persons from ISPs’s customers. 

Ed said that there were many PERSON objects referenced from assignments and that’s why he emailed the top 10 LIRs holding the most of these objects.

C. Geofeeds

Massimo Candela, NTT

Presentation available at:
https://ripe82.ripe.net/archives/video/608/

Harry Cross (Independent) asked if they thought of hosting the geofeed content inside the database instead of introducing another external link that could go wrong. He added that this would also bring the benefit of adding geolocation to WHOIS.

Massimo Candela said that this was probably not a good idea. He explained that having an external link is important because geofeed information is not something the RIPE NCC manages and that it’s the responsibility of the operator to maintain this link. Massimo added that it was also easier to host your own external link as you will avoid having access issues if someone else need to view the data.

Christian Bretterhofer (Independent) said that geofeed companies are asking for money while they get the data for free.

Massimo explained that if geolocation data is wrong you can also lose money and that geolocation providers accept correction via email for free.

D. Cloud Migration

Sander Buskens, RIPE NCC

Presentation available at:
https://ripe82.ripe.net/archives/video/609/

Randy Bush (RGnet) asked how AWS-dependent this project has become and how much effort was done to move to Google or to an European cloud provider.

Ed said that the RIPE NCC is careful to maintain their own in-house solutions and avoiding vendor lock-in.

Harry Cross (Independent) said that in other presentations about this topic, features inside the AWS ecosystem such as Fargate were mentioned. He asked if there has been some research done on using these features in relation with vendor lock-in.

Ed confirmed that the RIPE NCC was using Fargate as it simplifies development and there were using it as a standalone application on their server on-premises. He also mentioned that there was a conscious effort from the RIPE NCC to avoid vendor lock-in but mentioned that they will keep these concerns in mind for future plans.

Shane Kerr (NS1) asked if the goal of the database in the cloud was to get additional geographic locations.

Ed confirmed and added that the RIPE NCC was not looking for more service capacity but only to handle load. He added that the idea behind adding geographical location was to improve reliability.

Elvis Daniel Velea (V4Escrow) commented that it was a very bad idea to have the main database in AWS. He asked the RIPE NCC to reconsider.

Rüdiger Volk (Independent) asked how far-fetched the idea is to include the alternate targeting to an alternate (secondary) cloud provider in the test suite/process and if there was any preparation to allow load management.

Ed said that this mean that the RIPE NCC can test having an alternative cloud provider and added that the application is open source. Regarding load management, Ed said that the RIPE NCC was doing load testing to size its environment.

E. NWI - Open Proposals

Denis Walker, DB-WG Chair

Presentation available at:
https://ripe82.ripe.net/archives/video/610/

There were no questions.