Skip to content

Latest commit

 

History

History
73 lines (57 loc) · 3.07 KB

INPUTS.md

File metadata and controls

73 lines (57 loc) · 3.07 KB
Raw

Simulation Inputs

For some of the simulation experiments, we use traces from the The CAIDA Anonymized Internet Traces. In order to be able to download them, you need to request access to CAIDA.

Since the processing we do is not really part of the evaluation of Fancy, and to make your life easier, we provide you with the exact set of data inputs we use for our simulations. You can download them here or if you use the provided VM, you will already find the inputs at ~/fancy/fancy_sigcomm_inputs/.

Download and untar inputs

# to download
wget https://polybox.ethz.ch/index.php/s/w3To3lCCnwIPDlz/download -O fancy_sigcomm_inputs.tar.gz

# to uncompress
tar -xvf fancy_sigcomm_inputs.tar.gz

Inside the folder you will find the following folders.

fancy_sigcomm_inputs
├── equinix-chicago.dirB.20140619
├── equinix-nyc.dirA.20180419
├── equinix-nyc.dirB.20180816
├── equinix-nyc.dirB.20190117
└── zooming_info

For each of the four caida traces there is some simple precomputed files:

Global information computed over the entire 1h trace:

  1. <trace_name>.top: sorted list of all the /24 prefixes with the amount of total bytes sent in 1h and packets.
  2. <trace_name>.capinfos: capinfos of this trace
  3. <trace_name>.cdf: bytes and packets cdf per prefix.

Slice information. This contains trace information for a 30 second slice. You will just need the first slice number 0:

  1. <trace_name>_<slice_number>.bin: compressed version of the pcap. It just keeps basic five tuple info and timestamp.
  2. <trace_name>_<slice_number>.ts: for each prefix is has all the timestamps for every packet sent.
  3. <trace_name>_<slice_number>.cdf: bytes and packets cdf per prefix.
  4. <trace_name>_<slice_number>.freq: flows per second that are observed
  5. <trace_name>_<slice_number>.info: timestamp start and end (real times from the trace)
  6. <trace_name>_<slice_number>.rtts: All the RTTs we could infer from the trace, monitoring SYN and SYN ACKS.
  7. <trace_name>_<slice_number>_rtt_cdfs.txt: Compressed CDF of all the RTTs.
  8. <trace_name>_<slice_number>.top: All prefixes sorted by bytes and packets.
  9. <trace_name>_<slice_number>.dist: for each prefix and for each flow: start, end, duration, size, and rtt. This is used to generate flows.

Generating the inputs.

For a given 1h trace you can generate the input traces digests by using main_pcap_prefix_info from experiments/fancy/parse_traces/pcap_parse.py.

Download the following traces ['equinix-chicago.dirB.20140619', 'equinix-nyc.dirA.20180419', 'equinix-nyc.dirB.20180816','equinix-nyc.dirB.20190117']. And put them at traces_path.

Then run:

all_traces = [
    'equinix-chicago.dirB.20140619', 'equinix-nyc.dirA.20180419',
    'equinix-nyc.dirB.20180816', 'equinix-nyc.dirB.20190117']

main_pcap_prefix_info(
        traces_path, all_traces, slice_size=30, skip_after=30, slices=1,
        processes=1):