Topic: Scalability & Performance
DFINITY lead: @abk
Coordinator: @domwoe
Community Coordinator: @icme
Discord: Scalability
Meeting Notes: Link
Cadence: Monthly on the third Thursday of every month at 5:30pm CET
First meeting: July 21, 2022 at 5:30pm CET
The event is part of the regular calendar for technical working groups.
The Internet Computer provides a unique blockchain platform for decentralized applications. The actor model allows parallel execution of code in a given subnet and the IC can scale horizontally by adding more subnets. Already today, this allows the building of decentralized applications using smart contracts that aren’t possible on other blockchains. However, these are still early days for the Internet Computer, and design patterns for building decentralized applications on a global scale are only starting to emerge. The Internet Computer itself is also continuously maturing with performance improvements, new features, and better tooling. The goal of this working group is to foster a regular exchange between R&D at DFINITY and application developers with the following focus:
Updates on the relevant R&D roadmap and presentation of ideas
Collecting and discussing pain points of application developers
Identification and sharing of best practices and design patterns for building applications on the IC
Thanks a lot to @icme who provided the spark to initiate this working group. We hope that there’s interest in the community to participate in this working group. If you think we’re missing something important in the scope or you have ideas already for special topics you want to discuss or present please reply in this thread or the Discord channel.
For the first session of this new working group we’d like to start with an AMA on scalability to get a feeling about the questions and topics that you’re pondering with. Please give us a heads up by filling out the following Typeform.
Probably forgot quite a few (and can’t tag more than 10 people in a post as it turns out) so please tag others that might be interested or can contribute
Also would be great to get some people from the DFINITY Asia community & AstroX @zire@neeboo
Fee free to tag some people from the gaming community to understand their requirements going forward, as well as @rckprtr@hpeebles to understand their solutions & needs given that they are the top 2 apps on the IC (DSCVR/OpenChat)
Johan is on vacation this week. Hence, we need to postpone the discussion of the roadmap wrt. scalability and performance of the IC. In the survey, the most requested topic was “safe canister upgrades” and two 3/4 of the survey participants are using Rust for canister development.
One issue that devs encounter in canister upgrades is that the instruction limit gets reached while serializing or deserializing state to/from stable memory. One way to circumvent this is to use stable memory for storing state directly. Additionally, this allows also to use up to 8 GB for canister state currently.
There have been recent developments by Dfinity and the community to increase the usability of stable memory in Rust, and we’ll have two presentations on these:
I’ve created this Living Feedback Document for the members of the Scalability and Performance working group and those who are interested.
It contains viewable resource links and content with respect to what we’ve covered in previous meetings, and members who show up to the meeting will be permitted to comment/suggest questions or feedback to discuss in future meetings.
If you’d like to suggest a topic to be discussed in future meetings, you can do so by:
Commenting in this topic or the #scalability channel in the Developer discord
I noticed that the notes mention that Johann from DFINITY will soon work on Wasm multiple memories support. Has that been standardized by Wasm yet? (I know it’s available in wasmtime with a command-line flag.)
If that is shipped, won’t that deprecate most of the libraries that facilitate stable memory access, including ExperimentalStableMemory in Motoko? What is the ETA on this? Thanks.
Hey thanks for sharing !
I was late for the meeting so I would like to watch what I missed but I get this message when I try to access the recording : You cannot view this recording. No permission.
Multiple memories is still in Phase 3, i.e., not fully standardised. But it is already implemented in several engines, including Wasmtime, which the IC uses. Moreover, according to Andreas Rossberg, the proposal’s champion, and former principal engineer at DFINITY, the proposal has been unchanged for a long time and is not expected to change. Thus, we are rather comfortable experimenting with using it on the IC. New libraries for Rust and Motoko will be needed to take advantage of multiple memories. We haven’t started working on it in earnest, so it is too early to give an ETA. We’ll continuously update the Scalability & Performance WG with our progress. Best, Johan.
We moved this week’s (9/15) Scalability & Performance WG to next week 9/22 same time as usual. If you subscribed to the Technical Working Groups calendar the event should’ve been updated.
We know that there’s a lot of interest in discussing Inter Canister Query Calls (ICQCs) as well, and the plan is to have a discussion on this topic led by @ulan and @dsarlis in the next WG call in October.
)