As RPKI repository operator we noticed that a significant amount of traffic can be saved by using compression. However, due to the issues reported earlier (gzip bomb), compression was disabled by RPs.
By now there are multiple RPs (rpki-prover, rpki-client) that have implemented compression in either a separate process or a memory limited buffer. That way the total memory usage can not cause issues, while it speeds up retrieval and reduces the data volume for repositories.
I think this would be a good addition for routinator as well.
Rethinking RRDP fetching is part of the plan for 0.14, anyway, so this is good timing. I’ve added it to the milestone and we’ll consider it as one of the requirements for the rethink.
As RPKI repository operator we noticed that a significant amount of traffic can be saved by using compression. However, due to the issues reported earlier (gzip bomb), compression was disabled by RPs.
By now there are multiple RPs (
rpki-prover
,rpki-client
) that have implemented compression in either a separate process or a memory limited buffer. That way the total memory usage can not cause issues, while it speeds up retrieval and reduces the data volume for repositories.I think this would be a good addition for routinator as well.