Crates.io | route_verification_shared_struct |
lib.rs | route_verification_shared_struct |
version | 0.2.2 |
source | src |
created_at | 2023-11-06 01:13:52.727139 |
updated_at | 2024-10-16 03:46:42.007838 |
description | Parse RPSL in the IRR to verify observed BGP routes |
homepage | |
repository | https://github.com/SichangHe/internet_route_verification |
max_upload_size | |
id | 1026260 |
size | 12,086 |
RPSLyzer provides libraries and examples to
parse the Routing Policy Specification Language (RPSL)
from Internet Routing Registries (IRRs)
and verify interdomain routes from Border Gateway Protocol (BGP)
table dumps against them.
The focus of parsing is on the RPSL objects and attributes related to
routing policies, centering on the aut-num
objects.
For the verification,
we simply walk through the AS-path in each BGP route and
interpret the policy in each AS's aut-num
object with the context of
the route.
After parsing the RPSL, we expose an intermediate representation (IR) in a JSON-compatible nested Rust data structure. You may thus access this IR from other languages via the serialized JSON.
As a user, you need to install the required tools, set up the environments,
and use the route_verification
Rust package
directly.
Please refer to ./ARTIFACTS.md
for the our data acquisition, parsing,
verification, and analysis workflow.
We annotate the RPSL-related Request for Comments (RFCs) at ./rfcs/
.
Please check these documents for the RPSL-related terminologies we use and
the limitations of RPSLyzer.
The RPSL lexer at ./rpsl_lexer/
uses PyParsing and Python,
and is compatible with PyPy.
We publish it
on PyPI as rpsl-lexer
.
rpsl-lexer
tokenizes ("lexes")
specific RPSL syntaxes into an abstract syntax tree (AST),
especially mp-import
, mp-export
,
and the <peering>
and <filter>
portions they contain.
We primarily call this library via UNIX pipes from the Rust library that
lexes the RPSL (route_verification_lex
).
We chose to tokenize with PyParsing to leverage the power of parsing expression grammar (PEG) for recursively-defined expressions.
The RPSL parser, verification logic, and read-evaluate-print loop (REPL)
shell script examples at ./route_verification/
are a series of
Rust crates (Rust packages) and scripts.
All main library crates are re-exported in
the route_verification
crate.
All re-exported crates have route_verification_
prepended to
their path names. Among these crates:
route_verification_ir
defines the IR and the relevant procedures.
route_verification_lex
lexes the RPSL source code into the AST.
It requires setting up rpsl-lexer
for lexing (see above).
route_verification_parse
parses the RPSL source code into the IR.
It leverages route_verification_lex
for lexing,
then parses the AST into the IR.
route_verification_irr
parses the RPSL source code from IRRs and
merges them into a single IR.
route_verification_as_rel
parses Center for
Applied Internet Data Analysis (CAIDA)'s
AS-relationship
dataset
to augment the verification.
route_verification_bgp
optimizes the IR for querying and
verifies BGP routes against the IR.
It can optionally merge in information from
the AS-relationship dataset (via pseudo as-set
s),
and apply special cases based on
these relationships during verification.
Additionally,
route_verification
provides a command-line interface (CLI)
to parse IRRs and to test run verification.
The REPL scripts are in
the module at ./route_verification/src/evcxr_examples.rs
.
route_verification_rib_stats
is a main script to
generate statistics for all BGP table dumps at a directory, on the AS,
AS-pair, and route levels.
We chose Rust for the IR, parser,
and verification logic for its strongly-typed enum
eration and
satisfactory performance at CPU-bound tasks.
The scripts we use to analyze and
visualize the results at ./scripts/
serve as examples for these tasks.
Most of them are in Python and
leverages common Python data analysis libraries,
though one of the CPU-bound scripts is in Rust.
./ARTIFACTS.md
and ./ALIASES.md
explain how to run the scripts and
the meanings of common variable names.
In addition to the code and short documentation, this repository's Issues contain detailed discussions and development records about this project. We recommend using GitHub's search to find relevant information in case you encounter issues.
We leverage user-friendly, reproducible, and automatic build tools everywhere.
All Rust crates use Cargo,
and Python libraries and scripts use Rye. Please see ./ARTIFACTS.md
for
more information.
Enable logging:
export RUST_LOG=route_verification=trace
Enable backtrace (stack trace) in error messages:
export RUST_BACKTRACE=1
Please see the GitHub Actions at .github/workflows/
for up-to-date tests.
Maintenance mode. We are not actively developing this project. Only bug-fix contributions will be considered. Please fork and modify as needed.
Internet Route Verification Server is an abandoned attempt to store the IR and the verification results in a PostgreSQL database and serve predefined queries via REST APIs. We abandoned it due to its limited utility and the tebibytes of disk space needed.
This is the code and Issue repository corresponding to the paper: RPSLyzer: Characterization and Verification of Policies in Internet Routing Registries. This paper was accepted at ACM IMC'24. Please find the camera-ready version among the GitHub Releases.