sui-shared-objects is a Rust-based project that provides convenient tools to estimate and visualize how often transactions operate on shared objects on the Sui network, using the Sui Rust SDK. We also list popular dApps that utilize shared objects on Sui.
Sui is a layer-1 smart contract platform that utilizes an object-centric data model: the basic unit of storage in Sui is an object. The Sui ledger, therefore, stores a collection of programmable objects, each with a globally unique ID.
From the ownership point of view, there are two types of objects in Sui:
- Owned objects: owned by an address and can only be used by transactions signed by that owner address at a time.
- Shared objects: do not have a specific owner; anyone can read or write these objects.
Mutable owned objects are single-writers, and thus, transactions involving only owned objects may bypass consensus on sequencing in Sui. Mutable shared objects (multi-writers), however, require consensus to sequence (order) reads and writes. See Sui Lutris for detail.
Owned objects are the most common case in Sui. According to the (old) Sui documentation, many transactions (e.g., asset transfers, NFT minting, smart contract publishing) can be realized involving only owned objects.
Warning
This list of single-writer applications was once provided on the Sui
documentation at https://docs.sui.io/learn/single-writer-apps but removed
at some point later. If it cannot be found in the 21dad3e
commit
on the Sui repo, we archived it here.
On the other hand, the (old) Sui documentation also claims that many use cases require shared objects that can be manipulated by two or more addresses at once (e.g., an auction with open bidding, a central limit order book that accepts arbitrary trades). Thus, it is reasonable to investigate how often Sui transactions actually operate on shared objects.
This analysis of the Sui network may give insights into how frequently the use cases that require shared objects appear on the Sui smart contract platform. The interest in this analysis stems from the fact that transactions with shared object inputs require sequencing via the consensus protocol. Therefore, understanding how often apps involve shared objects and what those use cases are is one of the first key steps in improving the efficiency of object-based smart contract architectures. This analysis is relevant for smart contract platform designers and smart contract developers.
Before you start, please refer to the Sui Rust SDK documentation and/or the Sui Rust SDK source code if you need more information about Sui Rust SDK and available API methods it provides.
This project assumes Rust
and Cargo
are installed, and that there is an
available internet connection. Please refer to the
Rust and
Cargo
documentation for the installation instructions.
- Clone the project:
git clone https://github.com/roman1e2f5p8s/sui-shared-objects.git
cd sui-shared-objects
- Build the project:
cargo build --release
Building the project shall create three executable files:
query-txs
;metrcis
;query-obj
.
Tip
On UNIX-like systems, these can be executed using ./target/release/<NAME>
,
where <NAME>
is one of the three executables listed above.
Tip
On Windows, these can be executed using .\target\release\<NAME>.exe
,
where <NAME>
is one of the three executables listed above.
A more detailed description of how to use these executables and what they do are given in the next sub-sections.
Important
Use query-txs
to query all the transactions (i.e., programmable transaction
blocks) for a given epoch, and pre-process them to save only the relevant
data we need for this analysis.
For example,
./target/release/query-txs --epoch=0
will query all transactions for epoch 0
and pre-process them to collect
only the relevant data for this analysis, as specified here.
By default, the processed data will be saved in data/workspace1/
, one file
per epoch. You can create another workspace using the --workspace
command
line argument for query-txs
.
For more information and all command line arguments, use --help
:
./target/release/query-txs --help
See this specification for the description and explanation
of which data about Sui transactions query-txs
collects and stores.
Important
Use metrics
to calculate metrics of interest and obtain a set
of all shared object IDs for further analysis.
Specifically, executing
./target/release/metrics
will calculate metrics for all epochs the data is collected for using
query-txs
, collect IDs of all shared objects, and store them in json
files
whose structures are described here.
Note
By default, the metrics and shared object IDs will be saved in
results/workspace1/
. The workspace must be the same as specified in
query-txs
and can be changed using --workspace
command line argument
for metrics
.
For more information and other command line arguments, use --help
:
./target/release/metrics --help
Important
Use query-obj
to obtain information about a set of collected shared objects
(returned by metrics
) and packages that implement them:
./target/release/query-obj
Note
By default, the data shared objects and packages will be saved in
results/workspace1/
. The workspace must be the same as specified in
metrics
and can be changed using --workspace
command line argument
for query-obj
.
For more information and all command line arguments, use --help
:
./target/release/query-obj --help
See this specification for the description and
explanation data about shared objects and packages query-obj
collects and
stores.
Recall the following concepts from Sui:
- Epoch: In Sui, each epoch takes approximately 24 hours.
- Checkpoint: A checkpoint (also called sequence number) in Sui changes approximately every 2-3 second.
We also need to define auxiliary concepts:
- Interval: An interval is a period of time expressed in the number of checkpoints.
- Contention: Contention is a situation when multiple transactions touch the same shared object at the same time, i.e., concurrently access that shared object.
- Shared-object transaction: A shared-object transaction has at least one shared object in its inputs.
The following metrics are defined and calculated:
- Density: The density is the ratio of the number of shared-object transactions to the number of all transactions. The density is a number between 0 and 1; the higher the density, the more transactions operate on shared objects.
- Contention degree: The contention degree is the ratio of the number of shared-object transactions (within some interval) to the number of shared objects touched by those transactions (within the same interval). The contention degree is a number between 0 and ∞. A contention degree of 1 means that each shared-object transaction operates on a single different object, on average; values larger than 1 indicate multiple shared-object transactions contending for the same shared object; values smaller than 1 mean a transaction touches multiple shared objects, on average.
- Contended fraction: The contended fraction is the ratio of the number of shared objects (within some interval) touched by more than one transaction to the total number of shared objects (within the same interval). The contended fraction is a number between 0 and 1. The higher the contended fraction, the more shared objects are touched by more than one transaction.
We also calculate and plot the following simple metrics:
- The total number of transactions (per epoch).
- Number of shared-objects transactions (per epoch).
- Number of shared objects (touched per epoch).
TODO