twitter / cache-trace Goto Github PK

problem
The https://ftp.pdl.cmu.edu/pub/datasets/twemcacheWorkload/open_source url is no longer accessible (this used to work in the past).

questions

1. Whats the best way to download this dataset (in the US northeast)?
2. Could we get checksums of each file (e.g. a sha256)?

When I was downloading the traces, I noticed some files have a .sort tag and some do not. Is there much difference between a .sort file and one that is not? Please let me know the difference.

Hi,

Thanks for the exciting work and traces.

I wonder if you can recommend a tool to replay these traces.

Thank you so much for your insights and time!

Hi!

Note that during key anonymization, we preserve the namespaces, for example, if the anonymized key is nz:u:eeW511W3dcH3de3d15ec, the first two fields nz and u are namespaces, note that the namespaces are not necessarily delimited by :, different workloads use different delimiters with different number of namespaces.

I've already found several different special characters, and I have no idea whether they are the delimiters or not.
Can you share more details about namespace delimiters? like some regex?
It will help me a lot!

Thanks so much!!

Is it possible to put the uncompressed trace files on ftp.pdl.cmu.edu or other places? In this way i can use range http header to download part of the file and send to cache. For example the meta KV trace, I can use curl -v -r 0-5000 'https://cachelib-workload-sharing.s3-us-west-2.amazonaws.com/pub/kvcache/202206/kvcache_traces_1.csv' to get first 5000 bytes.

I want to do some research about caching with low locality(means large reuse distance), but I don't find such nice traces.

So I want to ask is there such kind of traces in this project?

Or any scripts to get L2 traces by doing a L1 filtering?

Thank you for your time.

A great work!
But I found the timestamps of the last queries in different clusters are different. So I'm wondering what's your sampling strategy?

Thanks!