{
  "slug": "cloud-disk-error-prediction",
  "title": "Azure disk-error prediction caught 36.5% of faulty disks at a 0.1% false-positive rate, and random cross-validation said 91.6%",
  "kind": "dataset",
  "summary": "Xu et al., USENIX ATC 2018, built a disk-error predictor for Azure. At a 0.1% false-positive rate it caught 36.50% of faulty disks on one test set (29.67% to 41.09% across three), against 15.51% to 30.51% for SVM and random forest baselines on SMART data; a random split gave 91.64% on the same data. Rows from Backblaze (2016) show the SMART gap in another fleet.",
  "license": "Small derived table of figures printed in Xu et al. (USENIX ATC 2018) and a Backblaze post of 6 October 2016, with credit. Not a Creative Commons license. The USENIX proceedings state that rights to individual papers remain with the author or the author's employer. This file is not a copy of the paper and not the data.",
  "licenseUrl": "https://www.usenix.org/system/files/conference/atc18/atc18-xu-yong.pdf",
  "sourceUrl": "https://www.usenix.org/system/files/conference/atc18/atc18-xu-yong.pdf",
  "accessed": "2026-10-10",
  "rowCount": 15,
  "columns": [
    {
      "name": "measure",
      "type": "string",
      "description": "The quantity as the source names it."
    },
    {
      "name": "value_low",
      "type": "number",
      "description": "Lower end of a printed range. Empty when the source prints a single value."
    },
    {
      "name": "value_high",
      "type": "number",
      "description": "Single printed value, or the upper end of a range. For 'more than half' it is the stated bound."
    },
    {
      "name": "unit",
      "type": "string",
      "description": "Unit of the value columns: disks, blocks, files, percent of disks, percent of mismatches, or probability."
    },
    {
      "name": "scope",
      "type": "string",
      "description": "Population and window the value applies to."
    },
    {
      "name": "note",
      "type": "string",
      "description": "What the value is not, or the source wording behind it."
    },
    {
      "name": "citation_id",
      "type": "string",
      "description": "Id of the opened source in the citations list."
    }
  ],
  "downloadPath": "/data/datasets/cloud-disk-error-prediction.csv",
  "humanPage": "https://hesela.com/analyses/cloud-disk-error-prediction/",
  "limits": [
    "One cloud system of one company, one month of training and one month of testing.",
    "Labels come from engineers' root-cause analyses and some disks return from repair.",
    "Only TPR, FPR, and AUC are reported; precision is not.",
    "The 63000 minutes saved is the authors' own average after deployment.",
    "The Backblaze rows measure SMART warnings in another fleet, not this model."
  ],
  "citations": [
    {
      "id": "xu-atc18-pdf",
      "title": "Improving Service Availability of Cloud Systems by Predicting Disk Error, Xu et al., USENIX ATC 2018 (USENIX PDF, pages 481 to 494)",
      "url": "https://www.usenix.org/system/files/conference/atc18/atc18-xu-yong.pdf",
      "accessed": "2026-10-10"
    },
    {
      "id": "backblaze-smart-2016",
      "title": "What SMART Stats Tell Us About Hard Drives, Backblaze, 6 October 2016",
      "url": "https://www.backblaze.com/blog/what-smart-stats-indicate-hard-drive-failures/",
      "accessed": "2026-10-10"
    }
  ]
}