id: databank
name: DataBank
domain: www.databank.com
kind: operator
mode: hybrid
priority: 1
license: "Publicly accessible operator pages; factual data only; © DataBank"
attribution: "DataBank"
homepage: https://www.databank.com/data-centers/
notes: "Facility pages /data-centers/<metro>/[<campus>/]<facility>/ publish code (ATL6), street address, IT sq ft and critical IT load. JSON-LD on these pages is the corporate HQ and is ignored."
coverage: { operators: ["DataBank"], countries: ["US", "GB"] }
fetch: { level: 1, maxLevel: 2, rpm: 20, concurrency: 2, respectRobots: true, maxCreditsPerRun: 50 }
discovery:
  sitemap: ["https://www.databank.com/db_data_center-sitemap.xml", "https://www.databank.com/db-press-release-sitemap.xml", "https://www.databank.com/db-news-sitemap.xml"]
  include: ["/data-centers/[a-z0-9-]+/", "/press-release", "/news/"]
  exclude: ["\\?", "#", "/data-centers/data-center-design/", "/data-centers-map/"]
  classify:
    - { pattern: "/data-centers/[a-z0-9-]+/[a-z0-9-]+-campus/$", group: facility_index, pageType: facility_index, priority: 40 }
    - { pattern: "/data-centers/[a-z0-9-]+/(?:[a-z0-9-]+/)?[a-z0-9-]+/$", group: facility_pages, pageType: facility_page, priority: 60 }
    - { pattern: "/data-centers/[a-z0-9-]+/$", group: facility_index, pageType: facility_index, priority: 40 }
    - { pattern: "/(press-release|news)", group: newsroom, pageType: press_release, priority: 50 }
schedule: { facility_pages: weekly, facility_index: monthly, newsroom: daily, sitemap: weekly }
extractors:
  facility: { match: ["/data-centers/[a-z0-9-]+/"], kind: facility, parser: databank_facility_v1 }
defaults: { operatorName: DataBank, facilityType: colocation }