Skip to main content
WeaviateLinkedIn

Weaviate discusses trade-offs between HNSW and HFresh for vector index selection in production systems.

Source: Autobound Signal API · 47 signals tracked for Weaviate · updated

Get Weaviate's live signals in Claude or ChatGPT, or by API

AskWhat changed at Weaviate this week? Use Autobound signals for weaviate.io.

Source

LinkedInSep 9, 2026
likes
53
comments
1

Post

Is the fastest vector index always the right vector index? HNSW is difficult to beat when an application needs the lowest latency and highest query throughput. But production systems optimise for different constraints: 1. Some need to support very large datasets without keeping the full index in memory. 2. Some are limited by available memory or infrastructure cost. 3. Some change frequently and cannot depend on disruptive full-index rebuilds. HFresh takes a different approach. It uses a compact HNSW index to route each query towards the most relevant centroids, then searches compressed postings stored on disk. Underneath that architecture: - RQ8 makes centroid vectors approximately four times smaller before overhead. - RQ1 reduces stored posting-vector data by up to 32 times compared with 32-bit floats. - Background split, merge, and reassign operations keep the index balanced incrementally. The result is a disk-based index designed for lower heap usage, controlled query I/O, and large mutable datasets - without periodic full rebuilds. The choice is not HFresh or HNSW in every situation. Choose HNSW when peak query performance is the priority. Choose HFresh when memory efficiency matters more than peak throughput. Read how HFresh works: https://lnkd.in/dgPHmXXg

linkedin.com/posts/weaviate-io_is-the-fastest-vector-index-always-the...Read the full source

Extracted by Autobound

From the Signal API record
Signal
LinkedIn

What this signalsCompany posts often show what the team is pushing right now.

More LinkedIn signals at other companies

The full record

From the Signal API record

Topics and mentions

Tags

  • Big Data
  • Database
  • Artificial Intelligence

Initiatives

  • selecting appropriate vector index based on constraints
  • optimizing vector index for memory efficiency

Pain points

  • large datasets exceeding available memory
  • memory limitations or infrastructure cost constraints
  • frequent data changes requiring disruptive index rebuilds

Technologies named

  • HNSW
  • HFresh

Extraction

Detected
Sep 9, 2026
signal_type
linkedin-post-company
signal_subtype
linkedinPost

Use this data

Get every LinkedIn signal for Weaviate and the companies you sell to, in the tools you already use.

  1. Ask Claude about it

    Connect Autobound to Claude, Claude Code or Cursor with MCP. Then ask: “What changed at Weaviate this week?”

  2. Send it to your own tools

    The Signal API returns LinkedIn signals for any list of companies as JSON, for your CRM, warehouse or app.

  3. Try it free

    Sign up and spend your free credits on the companies you sell to.

    Start Free1,000 free credits

The API returns more than this page shows

This page shows a preview. The full linkedin-post-company record in the Signal API and MCP can also have these 8 fields. Some fields are empty for some signals.

Company

  • linkedin_urlValue in the API
  • industriesValue in the API
  • employee_count_lowValue in the API
  • employee_count_highValue in the API
  • revenueValue in the API
  • descriptionValue in the API

Signal

  • signal_nameValue in the API
  • associationValue in the API
Show the full JSONThe record on this page and the API request

GET /v1/signals/3368c8d2-1e80-43c6-ab94-e24a87bf12f4 returns this record as JSON. POST /v1/companies/enrich returns every signal for weaviate.io.

{
  "signal_id": "3368c8d2-1e80-43c6-ab94-e24a87bf12f4",
  "signal_type": "linkedin-post-company",
  "signal_subtype": "linkedinPost",
  "detected_at": "2026-09-09T14:00:57.749+00:00",
  "company": {
    "name": "Weaviate",
    "domain": "weaviate.io"
  },
  "data": {
    "tags": [
      "Big Data",
      "Database",
      "Artificial Intelligence"
    ],
    "summary": "Weaviate discusses trade-offs between HNSW and HFresh for vector index selection in production systems.",
    "post_url": "https://www.linkedin.com/posts/weaviate-io_is-the-fastest-vector-index-always-the-right-activity-7503452392387145729-k01B",
    "num_likes": 53,
    "post_text": "Is the fastest vector index always the right vector index?\n\nHNSW is difficult to beat when an application needs the lowest latency and highest query throughput.\n\nBut production systems optimise for different constraints:\n\n1. Some need to support very large datasets without keeping the full index in memory.\n2. Some are limited by available memory or infrastructure cost.\n3. Some change frequently and cannot depend on disruptive full-index rebuilds.\n\nHFresh takes a different approach.\n\nIt uses a compact HNSW index to route each query towards the most relevant centroids, then searches compressed postings stored on disk.\n\nUnderneath that architecture:\n\n- RQ8 makes centroid vectors approximately four times smaller before overhead.\n- RQ1 reduces stored posting-vector data by up to 32 times compared with 32-bit floats.\n- Background split, merge, and reassign operations keep the index balanced incrementally.\n\nThe result is a disk-based index designed for lower heap usage, controlled query I/O, and large mutable datasets - without periodic full rebuilds.\n\nThe choice is not HFresh or HNSW in every situation.\n\nChoose HNSW when peak query performance is the priority. Choose HFresh when memory efficiency matters more than peak throughput.\n\nRead how HFresh works: https://lnkd.in/dgPHmXXg",
    "initiatives": [
      {
        "topic": "selecting appropriate vector index based on constraints",
        "urgency": 0.7
      },
      {
        "topic": "optimizing vector index for memory efficiency",
        "urgency": 0.7
      }
    ],
    "pain_points": [
      {
        "topic": "large datasets exceeding available memory",
        "intensity": 0.6
      },
      {
        "topic": "memory limitations or infrastructure cost constraints",
        "intensity": 0.5
      },
      {
        "topic": "frequent data changes requiring disruptive index rebuilds",
        "intensity": 0.5
      }
    ],
    "posted_date": "2026-09-09T14:00:57.749Z",
    "num_comments": 1,
    "technologies_mentioned": [
      {
        "name": "HNSW",
        "status": "using"
      },
      {
        "name": "HFresh",
        "status": "using"
      }
    ]
  }
}

Long text fields are shortened on this page.

Looking up one signal by its id is free. Enrich costs 2 credits per signal returned; a call with no results is free.