模式支 持存储计算分离 高性能 基于 Rust 开发的分布式存储引 擎及图计算引擎，精细的内存 管理设计，内置索引系统，支 持毫秒级的并发查询响应速度 易用 AQL(Atlas Graph Query Language) ，类 SQL 的图查询 语言，内置上百种分析函数， 面向分析师友好，拥抱标准， 基于 openCypher 向 ISO GQL 迈进 实时大图 支持万亿节点存储及流式计算 X86/ARM - 基于 RUST 语言保证性能优势 - 分布式架构性能可线性扩展 - 针对大规模图优化的存算引擎 - 配合 Atlas 图平台，实现无代码图分析 - Query 性能分析模块，启发式提示优化 - 内置多种分析函数，面向分析师友好 -MVOCC 保证事务一致性 - 多副本管理保证数据服务高可用 - 在线备份提供容灾保障 高速 易用 可靠 Why Program API Analysis rules Optimization rules: MATCH -> pushdown Planning Strategies Graph Execution Code Gen JIT Cost model Storage backend （ Property Graph KV store ） Pushdown filter Cypher

including some standard ones: an event loop, portable access to the system execution context, nursery for spawned work4 P2300: STD::EXECUTION Proposes: • A set of concepts that represent: • A handle to a compute Example 1: Launching concurrent work6 EXAMPLE: LAUNCHING CONCURRENT WORK namespace ex = std::execution; int compute_intensive(int); int main() { unifex::static_thread_pool pool{8}; ex::scheduler concurrently on a custom execution context libunifex: https://github.com/facebookexperimental/libunifex7 EXAMPLE: LAUNCHING CONCURRENT WORK namespace ex = std::execution; int compute_intensive(int);

float* y) { for (int i = 0; i < n; i++) y[i] = x[i] + y[i]; } 35 |Execution Space Specifiers The __global__ execution space specifier declares a function as being a kernel. Such a function is: is: • Executed on the device, • Callable from the host. The __device__ execution space specifier declares a function that is: • Executed on the device, • Callable from the device only. https://docs __host__ execution space specifier declares a function that is: • Executed on the host, • Callable from the host only. It is equivalent to declare a function with only the __host__ execution space specifier

© 2018 Percona. 6 Why Not ? Generating Full Query Logs can be Expensive Logs can Consume a lot of Space Logs can be expensive to query © 2018 Percona. 7 Clickhouse Answers • 10x+ Agent https://github.com/honeycombio/honeytail Nice and Simple Self Contained GoLang Application Query Normalization Built-in Supported Parsers Meet HoneyTail © 2018 Percona. 13 https://github.co clicktail start • To run as a service © 2018 Percona. 15 MySQL Logs Primer General Query Log Binary Log Slow Query Log Audit Log © 2018 Percona. 16 MySQL Audit Logs to ClickHouse © 2018 Percona

the caller, and suspends the coroutine, subsequently calling the coroutine again continues its execution  co_return: returns from a coroutine (just return is not allowed)16 Coroutines  What are coroutines Can be used with condition_variable_any  std::stop_source  Used to request a thread to stop execution  Stop requests are visible to all associated stop_sources and stop_tokens  std::stop_callback calculate the midpoint of two numbers  lerp() to do linear interpolation  New unsequenced_policy (execution::unseq): algorithm is allowed to be vectorized84 Agenda  Modules  Ranges  Coroutines 

配置原则 服务优化 服务优化 公共选项 公共选项 选项 缺省值 推荐值 说明 max_connections 100 1024 MySQL 服务器同时处理的数据库连接的最大 数量 query_cache_size 0 ( 不打开 ） 128M 查询缓存区的最大长度，按照当前需求，一 倍一倍增加，本选项比较重要 sort_buffer_size 512K 128M 每个线程的排序缓存大小，一般按照内存可 里使用了 MySQL MySQL 部分自带函数，索引将失效，同时将无 部分自带函数，索引将失效，同时将无 法 法 使用 使用 MySQL MySQL 的 的 Query Cache Query Cache ，比如 ，比如 LEFT(), SUBSTR(), LEFT(), SUBSTR(), TO_DAYS() TO_DAYS() DATE_FORMAT() 执行情况，是否锁表，查看相应的 执行情况，是否锁表，查看相应的 SQL SQL 语句 语句  设置 设置 my.cnf my.cnf 中的 中的 long-query-time long-query-time 和 和 log-slow-queries log-slow-queries 能 能 够 够 记录服务器那些 记录服务器那些 SQL SQL 执行速度比较慢

Await Tree 的维护 Await Tree 的设计原理与实现 • Select First Poll Await Tree 的维护 Await Tree 的设计原理与实现 • Query 返回 Pending Await Tree 的维护 Await Tree 的设计原理与实现 • Select 返回 Pending Await Tree 的维护 Await Tree 的设计原理与实现 Tree 的设计原理与实现 • Select Ready Await Tree 的维护 Await Tree 的设计原理与实现 • Query Next Poll Await Tree 的维护 Await Tree 的设计原理与实现 • Query -> Handle Ready • 另： Cancellation Await Tree 的维护 Await Tree 的设计原理与实现

New(obj); // 0x9200 -> 0x7800 Sys_Query( ); Handle.Free( ); // Collection occurs. // Copy, compacting, etc. // obj now at 0x6800. // 0x9200 -> 0x6800 Sys_Query(hnd); hnd // Collection occurs occurs. // obj now at 0x7200. // 0x9200 –> 0x???? Sys_Query(hnd); // Collection occurs. // obj now “free”. // 0x9200 -> 0x7200 Sys_Done(hnd); hnd This is only one possible sequence.Memory model – Non-Reference

App Open Data App Centralization 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 App Centralization Decentralized 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

app size Threading ⛓ • NativeScript = single threaded, by default ? • Samples for background execution are now available. • WebWorkers API coming soon Angular ? • Angular 2 final—Now 100% more a