{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# NAS Drives: CMR/SMR, Failure Rates & Specs\n",
    "\n",
    "Exploring the dataset of 140+ NAS and enterprise hard drives - CMR vs SMR classification, real-world failure rates from Backblaze Drive Stats, and full specs."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "\n",
    "df = pd.read_csv('/kaggle/input/datasets/nettrust/nas-drives-cmrsmr-failure-rates-and-specs-2026/drives-2026-06-17.csv')\n",
    "print(f\"{len(df)} drives | {df['brand'].nunique()} brands | {df['line'].nunique()} product lines\")\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## CMR vs SMR breakdown\n",
    "\n",
    "SMR drives use shingled recording to pack more capacity - but they stall under sustained writes of a RAID rebuild, which can drop them from the array mid-rebuild."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "df['recording_tech'].value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## SMR drives to avoid in a NAS/RAID array"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "df[df['recording_tech'] == 'smr'][['brand', 'line', 'model', 'capacity_tb']].sort_values(['brand', 'capacity_tb'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Most reliable drives (Backblaze annualized failure rate, lowest first)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "df[df['afr_pct'].notna()][['brand', 'line', 'model', 'capacity_tb', 'recording_tech', 'afr_pct', 'reliability_drive_count']].sort_values('afr_pct')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data & API\n",
    "\n",
    "Full dataset, always up to date: [nasdisks.com/data/](https://www.nasdisks.com/data/)\n",
    "\n",
    "License: CC BY 4.0 for specs and CMR/SMR. Failure rates derived from [Backblaze Drive Stats](https://www.backblaze.com/cloud-storage/resources/hard-drive-test-data) - credit Backblaze when using."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "name": "python",
   "version": "3.12.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}