DOCSEARCHbyAlgolia
Submit Docs

Samzong's Technical Blog

website·Indexed 32 days ago

https://samzong.me

A collection of technical notes and tutorials covering Python development (FastAPI, Flask, Poetry, pyenv), Kubernetes (kubeadm, CKA preparation), Docker on Apple Silicon M1, SQL, and various DevOps tools like GitLab, Metabase, and MkDocs.

Tokens
106.6K
Snippets
659
Records
847
Agent score
56%

What's inside Samzong Blog

  1. Overview of ClickHouse for Large-Scale Data Retrieval

    ClickHouse is a column-oriented DBMS designed for Online Analytical Processing (OLAP) and high-performance fast read operations on large-scale data warehouses. Key architectural features that enable fast retrieval include:

    • Columnar Storage: Optimizes aggregation queries by reading only required columns.
    • Data Compression: Reduces physical storage and improves I/O efficiency.
    • Vectorized Query Execution: Processes multiple data values simultaneously to increase computation speed.
    • Materialized Views: Stores pre-computed aggregation data for instant retrieval.

  2. Overview of Canal for MySQL incremental data synchronization

    Canal is a tool designed to provide incremental data subscription and consumption by parsing MySQL database binary logs. It is commonly used for database mirroring, real-time backups, index construction/maintenance (such as inverted indexes), business cache refreshing, and incremental data processing with business logic. It supports MySQL versions 5.1.x, 5.5.x, 5.6.x, 5.7.x, and 8.0.x.

  3. Overview of Skoala-gateway capabilities

    Skoala-gateway is a distributed gateway middleware for north-south traffic governance within the DCE 5.0 cloud-native platform. Built on Contour and Envoy, it is compatible with Kubernetes and OpenShift. Key capabilities include:

    • API Management and routing forwarding
    • Interface rate limiting
    • Security authentication strategies
    • Blacklist/Whitelist management
    • MockAPI support
    • Multi-instance management for isolation across different clusters and projects

  4. Overview of MariaDB Galera Cluster

    MariaDB Galera Cluster is a multi-master system architecture implemented on the MySQL InnoDB storage engine that provides real-time data synchronization. Unlike master-slave setups, it is a 'dual-active' (multi-master) mode where every node can handle read and write operations. Key features include:

    • True multi-master architecture: any node can be read from or written to.
    • Synchronous replication: no lag between nodes and no data loss during node failure.
    • Tight coupling: all nodes maintain the same state.
    • Automatic node configuration: no need for manual backups when adding new nodes.

  5. Overview of the Linux Virtual File System (VFS)

    The Linux VFS (Virtual File System) is an abstraction layer that allows different types of filesystems (Network, Disk, and Special/Pseudo filesystems) to coexist and be accessed via a common interface. It abstracts four basic objects:

    1. Superblock Object: Represents an installed filesystem.
    2. Inode Object: Represents a specific file (the d_inode is the basis for hard links).
    3. Dentry Object (Directory Entry): Represents a directory component (e.g., in /dev/input/event5, /, dev/, input/, and event5 are all dentries). VFS uses a dcache to speed up path resolution.
    4. File Object: Represents a file opened by a process.

    This architecture enables system calls like open() and read() to work transparently across different physical or virtual storage backends.

  6. Overview of pt-query-digest

    pt-query-digest is a tool used to analyze MySQL slow queries. It can process data from binlogs, General logs, slow logs, SHOW PROCESSLIST output, or MySQL protocol data captured via tcpdump. The tool parameterizes query conditions and groups them to provide statistics on execution time, call frequency, and percentage of total load, helping developers identify and optimize problematic queries.

  7. Overview of Fluid data abstraction

    Fluid provides elastic data abstraction and acceleration for BigData and AI applications running in cloud environments. It allows for efficient data management and acceleration, often integrating with systems like Alluxio.

  8. Overview of Apache Dubbo

    Apache Dubbo is a cloud-native microservices framework based on RPC (Remote Procedure Call) development. It provides built-in capabilities for remote address discovery (via registry centers) and communication between microservices. Key features include service discovery, load balancing, and traffic scheduling, all of which are highly extensible.

  9. Overview of Alfred and Workflows for macOS

    Alfred is a productivity application for macOS designed to increase efficiency through powerful built-in features and extensions. A core component of Alfred (introduced in version 2.0) is the 'Workflow' system. A Workflow allows users to chain multiple steps of a task together using a set of rules, executing them sequentially to automate repetitive daily work processes.

  10. Overview of LLM-D Distributed Inference Framework

    LLM-D is a Kubernetes-native distributed inference framework developed by Red Hat, CoreWeave, Google Cloud, NVIDIA, and IBM Research. It aims to standardize production-grade AI inference by addressing high costs and latency. The core innovation is the separation of the two distinct phases of LLM inference: Prefill (compute-bound, high FLOPS requirement) and Decode (memory-bound, high bandwidth requirement), allowing them to be deployed on hardware optimized for their specific needs.

  11. Overview of NVIDIA Dynamo Inference Framework

    NVIDIA Dynamo is a high-throughput, low-latency inference framework designed for generative AI. Rather than replacing runtimes like TensorRT-LLM, vLLM, or SGLang, it acts as an intelligent scheduling layer. It optimizes distributed cluster efficiency through disaggregated inference (separating prefill and decode phases), KV-aware intelligent routing, dynamic GPU scheduling, and NIXL high-speed data transfer. The core is implemented in Rust for performance and safety, with a Python layer for extensibility.

  12. Install MongoDB on CentOS 6/7 using Yum

    To install MongoDB on CentOS 6/7 or RedHat-based systems, first add the appropriate repository file to /etc/yum.repos.d/ based on your version and architecture, then use Yum to install the mongodb-org meta-package which includes the server, shell, and tools.
    # Clean and update yum cache
yum clean all
yum makecache

# Install MongoDB
yum install mongodb-org