Repository navigation
feat(approach): Private Read - #218
Darkjay123 wants to merge 4 commits into
Conversation
Fills the Approach TBD gap on private-read (issue ethsystems#138). Adds the approach card comparing institution-run nodes, TEE-hosted private RPC and obfuscated aggregate reads; links the use-case card and adds the changelog entry.
rymnc
left a comment
There was a problem hiding this comment.
thank you for your contribution! comments inline
|
|
||
| ### Scenario | ||
|
|
||
| An asset manager monitors twenty addresses and queries transaction history ahead of execution. Every JSON-RPC call it makes is visible to its RPC provider, and the provider's node peers see the request pattern too. Read patterns leak portfolio composition, counterparty relationships and imminent trading intent, which MEV searchers monetize and competitors harvest. The institution does not want to run its own infrastructure everywhere, and the same read path must stay private against both the RPC provider and network observers, with regulator access available as scoped disclosure rather than standing visibility. |
There was a problem hiding this comment.
the eth wire protocol has no message for a balance, storage, call or log query. so the provider's node answers reads locally, and its peers never see them. only transactions that the node broadcasts reach peers.
suggestion:
| An asset manager monitors twenty addresses and queries transaction history ahead of execution. Every JSON-RPC call it makes is visible to its RPC provider, and the provider's node peers see the request pattern too. Read patterns leak portfolio composition, counterparty relationships and imminent trading intent, which MEV searchers monetize and competitors harvest. The institution does not want to run its own infrastructure everywhere, and the same read path must stay private against both the RPC provider and network observers, with regulator access available as scoped disclosure rather than standing visibility. | |
| An asset manager monitors twenty addresses and queries transaction history ahead of execution. Every JSON-RPC call it makes is visible to its RPC provider. Read patterns leak portfolio composition, counterparty relationships and imminent trading intent, which MEV searchers monetize and competitors harvest. The institution does not want to run its own infrastructure everywhere, and the same read path must stay private against both the RPC provider and network observers, with regulator access available as scoped disclosure rather than standing visibility. |
| - Local state is only as fresh as the node's sync | ||
|
|
||
| **Threat model:** | ||
| - Network observers correlate the node's traffic with known addresses |
There was a problem hiding this comment.
the eth wire protocol carries blocks and transactions, and it has no message that names an address to read. so the p2p traffic of a full node does not depend on which addresses it reads. the real leak of a self-hosted node is on the write path, where its transactions leave from its own ip. rfp-private-reads and pattern-tee-network-anonymity both say that a local node removes the provider exposure for reads.
this threat is also the reason for p: partial at line 45 and line 71. but line 172 calls this sub-approach the "strongest privacy".
suggestion:
| - Network observers correlate the node's traffic with known addresses | |
| - Transactions broadcast from the node reveal its IP as their origin; pair with [private transaction broadcasting](../patterns/pattern-private-transaction-broadcasting.md) |
| uses_patterns: [pattern-network-anonymity] | ||
| ``` | ||
|
|
||
| **Summary:** The institution runs its own full or archive node and queries local state, so no third party ever sees its reads. |
There was a problem hiding this comment.
the execution-apis spec has no method that returns the transactions of an address. the scenario needs that query. a node alone does not answer it. it needs a self-hosted indexer. a desk that runs a node but calls a third-party indexer or explorer for history still leaks those reads.
suggestion: add the self-hosted indexer to how it works, and add "history queries to third-party indexers" to the threat model. i did not check every client's indexing features.
| uses_patterns: [pattern-tee-based-privacy, pattern-tee-key-manager] | ||
| ``` | ||
|
|
||
| **Summary:** Queries are encrypted to an attested enclave operated by a private-RPC provider, which serves state without observing what is read or by whom. |
There was a problem hiding this comment.
line 84 says that institutions authenticate to the provider. so the provider knows who asks, and when, and how often. the use case lists "query source" as the first must-hide item at line 28.
the enclave also does not hide which state it reads. pattern-tee-based-privacy says that the host controls every byte that crosses the enclave boundary. state lookups from the enclave are visible to the host unless the store is oblivious. rfp-private-reads lists oram for this reason.
suggestion:
| **Summary:** Queries are encrypted to an attested enclave operated by a private-RPC provider, which serves state without observing what is read or by whom. | |
| **Summary:** Queries are encrypted to an attested enclave operated by a private-RPC provider. The provider cannot read query content, but it sees who queries, when and how much, and it needs an oblivious state store (ORAM) to hide which state the enclave reads. |
| ```yaml | ||
| maturity: documented | ||
| context: both | ||
| crops: { cr: medium, o: partial, p: full, s: medium } |
There was a problem hiding this comment.
this scores the tee rpc above pattern-tee-based-privacy, which is cr: low and p: partial. the private relay sub-approach, a provider-gated rpc, is also cr: low. but the pure tee rows in private payments and private mmf use cr: medium and p: full, with o: no and s: low. so master is not consistent either. also, the provider authenticates and bills each caller. so p: full does not hold under the rubric.
suggestion: pick one score set for a single-operator tee, give the reason, and use it here and in #217. then update the matrix row and the rationale.
| related_use_cases: [private-oracles] | ||
|
|
||
| primary_patterns: [pattern-tee-based-privacy, pattern-private-information-retrieval, pattern-network-anonymity] | ||
| supporting_patterns: [pattern-tee-key-manager] |
There was a problem hiding this comment.
pattern-tee-key-manager covers custody of institutional signing keys under a signing policy. this card uses it for query decryption keys, which that pattern does not cover.
pattern-tee-network-anonymity is missing. its example is almost this card's scenario: a fund manager reads fifty token balances without the rpc provider learning them. the card body has no links. so open question 2 at line 187, about reads and broadcasts, does not link the broadcasting approach.
suggestion: drop pattern-tee-key-manager here and at line 79. add pattern-tee-network-anonymity and pattern-private-transaction-broadcasting, and link the patterns in the body.
|
|
||
| **Summary:** Read patterns are hidden inside aggregate traffic and cryptographic retrieval: batching, timing randomization and PIR-style state access make individual reads unattributable. | ||
|
|
||
| **How it works:** Reads from many consumers are batched at a gateway, and timing is randomized within windows so request correlation fails. For specific record lookups, PIR-style retrieval lets the consumer read a value the gateway cannot identify. The gateway is designed so that its own logs are structurally unable to answer "who read what", rather than protected by policy alone. |
There was a problem hiding this comment.
this line batches reads from many consumers at the gateway. so the gateway receives each request, and it sees who sent it and when. batching and timing jitter hide reads from the systems behind the gateway. they do not hide them from the gateway. only the pir part hides the target record from the gateway. so the claim that the gateway logs cannot answer "who read what" holds only for pir reads.
line 122 also says that pir parameters resist traffic analysis. pattern-private-information-retrieval says that network metadata is out of scope for pir.
suggestion: say that the gateway sees the sender and timing of each request. say that pir hides only which record is read. consider a split into "pir over replicated state" and "network-anonymized rpc". each one then has one adversary.
| - A single consumer's volume dominates the batch, re-identifying it | ||
| - Strict real-time latency rules out batching windows | ||
|
|
||
| CROPS rationale: CR high, because reads are hidden within aggregate traffic without operator permission, though the operator still gates admission. O partial: the design is open, but working deployments are operator-specific. P partial: batch membership is hidden while timing and volume leak by design. S medium: batch discipline and honest gateways carry the privacy. |
There was a problem hiding this comment.
"working deployments are operator-specific" conflicts with maturity: concept and with "there is no standard implementation" at line 162. the cost class "low" at line 150 also conflicts with pattern-private-information-retrieval. the pattern reports seconds per query at 2^21 records and minutes at 2^25.
suggestion: write "no deployments" here, and set the cost class to "high (server compute)" or "not measured".
| Covered by [Private Read](../approaches/approach-private-read.md), which compares institution-run nodes, TEE-hosted private RPC and obfuscated aggregate reads, with a 10-row comparison matrix, persona perspectives and a conditional recommendation. Consider: | ||
| - Private RPC services with query obfuscation | ||
| - Query batching and timing randomization | ||
| - Institution-run infrastructure where feasible |
There was a problem hiding this comment.
the edit keeps "consider:" and the old bullet list under "covered by". it also describes the card format ("10-row comparison matrix"). #215 replaced the list with the recommendation at private-messaging.md line 55.
suggestion:
| Covered by [Private Read](../approaches/approach-private-read.md), which compares institution-run nodes, TEE-hosted private RPC and obfuscated aggregate reads, with a 10-row comparison matrix, persona perspectives and a conditional recommendation. Consider: | |
| - Private RPC services with query obfuscation | |
| - Query batching and timing randomization | |
| - Institution-run infrastructure where feasible | |
| See [Approach: Private Read](../approaches/approach-private-read.md): institution-run nodes where the institution can operate them, a TEE-hosted private RPC otherwise, and obfuscated aggregate reads as a research path. |
|
|
||
| ## [Unreleased] | ||
|
|
||
| - feat(approach): [Private Read](approaches/approach-private-read.md), filling the "Approach TBD" gap on [Private Read](use-cases/private-read.md), whose recommended-approaches section now links to the card. Compares institution-run nodes (production, I2I+I2U), TEE-hosted private RPC (documented) and obfuscated aggregate reads (concept), with a 10-row comparison matrix, persona perspectives, a conditional recommendation and open questions. |
There was a problem hiding this comment.
the entry has no pr link. .claude/CLAUDE.md asks for one.
suggestion:
| - feat(approach): [Private Read](approaches/approach-private-read.md), filling the "Approach TBD" gap on [Private Read](use-cases/private-read.md), whose recommended-approaches section now links to the card. Compares institution-run nodes (production, I2I+I2U), TEE-hosted private RPC (documented) and obfuscated aggregate reads (concept), with a 10-row comparison matrix, persona perspectives, a conditional recommendation and open questions. | |
| - feat(approach): [Private Read](approaches/approach-private-read.md), filling the "Approach TBD" gap on [Private Read](use-cases/private-read.md), whose recommended-approaches section now links to the card. Compares institution-run nodes (production, I2I+I2U), TEE-hosted private RPC (documented) and obfuscated aggregate reads (concept), with a 10-row comparison matrix, persona perspectives, a conditional recommendation and open questions. ([#218](https://github.com/ethsystems/map/pull/218)) |
- PIR sub-approach: maturity concept and the pattern's own scores (CR low single-operator case, O yes, S high); rationale and matrix updated - batching economics stated: server touches every bit per query, 121 MB hint per 1 GB dataset, update cost binds for price data - threat model: timing and rate leak trading intent; non-oblivious implementations may leak the index - integrity: pair PIR with a signed snapshot or state-root anchor; drop the avoid-when bullet that ruled PIR out of every oracle use - TEE sub-approach: operator still sees which consumer connects and when; drop pattern-tee-key-manager (signing-key misuse, same as ethsystems#218); add pattern-network-anonymity for the source; single-operator score set CR low/O partial/P partial/S medium - batched delivery: pattern-commit-and-prove replaced with pattern-mixnet-anonymity; pattern-zk-tls dropped from frontmatter; disclosure-keys pattern added for scoped regulator access - cross-provider aggregation moved to an open question - default and decision-factor summaries aligned with the review's suggested wording - changelog entry gains its PR link
- scenario: peers never see RPC reads; the eth wire protocol carries no read messages - institution-run nodes: transaction history needs a self-hosted indexer; third-party indexer queries added to the threat model; broadcasts reveal the node IP, paired with pattern-private-transaction-broadcasting - TEE-hosted RPC: provider sees who queries, when and how much; ORAM needed to hide which state the enclave reads; drop pattern-tee-key-manager (signing-key misuse); add pattern-tee-network-anonymity and link it - single-operator TEE score set (CR low/O partial/P partial/S medium), consistent with ethsystems#217; rationale and matrix updated - obfuscated aggregate reads: gateway sees sender and timing; PIR hides only the record; honest no-deployments maturity and high server-compute cost - use-case recommended-approaches section rewritten per review; changelog entry gains its PR link
# Conflicts: # CHANGELOG.md
Fills the "Approach TBD" gap on the Private Read use case from issue #138.
What this adds
New approach card Private Read (
approaches/approach-private-read.md) comparing three sub-approaches:Each sub-approach carries CROPS scores with written justification; the card follows the v2 template (issue #151) with the 10-row comparison matrix, persona perspectives, a conditional recommendation and open questions.
Accompanying changes
use-cases/private-read.md: the "Approach TBD" recommended-approaches section now links the card.CHANGELOG.md:[Unreleased]entry in the style of the recent approach PRs.Checks
All six required v2 sections present, title prefix and frontmatter complete, body word count within the documented approach limit. The Vale prose gate runs in CI.
Part of #138.