Performance Lets dive into Performance issues. • Everything in JavaScript defaults to being on the same thread. Too much work on main thread • Android nested layouts • Functions and objects defined

Third-party support (Javascript better than Dart) 利用 Rust 重塑移动应用开发 跨平台开发的优势和局 限性 Cons: - Performance - Native feature utilize - Existing Codebase 跨平台开发到到底 应该跨什么？ UI or Logic ? 利用 Rust 重塑移动应用开发 的特点 Why Rust? - Cross platform - Performance - Thread Safe - Memory Safe - Love 利用 Rust 重塑移动应用开发 Rust 在移动应用 开发中的应用 Why Rust? - Cross platform - Performance - Memory Safe - Love 利用 Rust 重塑移动应用开发 hardware wallet for the Web3 world 利用 Rust 重塑移动应用开发 Photo / image / chart 采用 Rust 重构业务逻 辑的背景和动机 Performance The same business logic will run on different platforms such as embedded devices, IOS, Android

Centralized Decentralized Performance High High, less than Web2 Query Capability Powerful Powerful, less than Web2 Open Source Code Open or Not, mostly NOT Open Data Resiliency Normal High Data Openness Closed Decentralized Underlying Nodes Many Few Performance Low High Query Capability Low High Open Source Code Open Open Hot Data Capacity Low High Data Resiliency High High Data Openness Open Open Trustlessness

Raw Text Log Files High Compression (column store + LZ4) • Typically 100x faster than MySQL on Single Server • Semi-Linear Scalability to 100s of nodes High Performance © 2018 Percona. Query Logs to ClickHouse © 2018 Percona. 23 When to use Slow Query Logs ? Everything Else  Performance Analyses Debugging © 2018 Percona. 24 Slow Log Record Example © 2018 Percona. 25 Configuring

shutdown_resource_X() No problem - No use of unavailable resource Concurrency TS2 Global Cleanup Performance? Simple and strong semantics Hazard Pointer Synchronous Reclamation Beyond Concurrency TS2 – Maged objects retired to it are destroyed before the the completion of destruction of Container. • Problem: High setup overhead of constructing/destroying per custom domain hazard pointers. • Even worse if many 2018: Global cleanup deprecated. Still supported for testing and microbenchmarking. • 2020: One user: High frequency of retiring objects. Unsharded list grew out of control. • 2020: Sharded cohorts without

−CVE-2019-14271 Docker CP vulnerability Pod Isolation Challenges • Noisy neighbor −Impact performance on CPU, Memory, Bandwidth, Buffer IO, PIDs, File descriptors # kubectl run --rm -it bb --image=busybox emulation • Flexible: Easy to customize to fit various network, storage and OS environments • High-performance: Leverage KVM capability Advantage of Rust-VMM • Google implements the first Rust-based light Rust-VMM Overview Rust Based • New language with a focus on correctness and performance • Compiled to native code offering performance similar to C • Memory management without garbage collection • Designed

Database + Caching + Data Subscription + Stream Processing Shell: Command Line Interface High Performance, Cloud Native Simplified Solution for Time-Series Data TDengine Telegraf Kafka OPC-UA Protocols • Streaming Pipeline • Transformer • Data Share • Visualization • Analysis Platform • High Avalibility 2022.12 2023.05 2023.09 Usability Functionality taosX - 集群运维 • 数据库复制 • 全量 / 增量备份

Guide back to agile principles • Servant Leader – Serve the team, not you • Performance Guardian – Facilitate high performance • Mission-Driven Leader – Serve the intentTransform Yourself • Detach from

platform is good for everything. Fast inner loop Makes UX easy Has tooling for workload High performance Avoids costly abstractions What the vendor providesA quick note on the details in this talk

request of Pod. However, in many cases, some nodes have low resource requests but high load, while some nodes have high resource requests but low load. Dynamic-Scheduler Node1 Node2 Kube-scheduler 5)*10+(1-0.4)*10=11 (1-0.5)*10+(1-0.6)*10=9 Assigned to Node1 DeScheduler If the pod is enabled high-load automatic migration Node2 Real Load Level Pod1 Pod2 Pod3 Node3 Real Load Level Pod2 Node1 Node1 Real Load Level Pod1 Pod1 migrate to Node1 Pod2 migrate to Node3 Descheduler Node2 load is high How to expand horizontally more quickly and flexibly HPA-controller Supporting horizontal scaling