Academic Journal

Search at Scale: Quantifying the Performance Trade-offs of ANNS on Milvus.

Bibliographic Details
Title: Search at Scale: Quantifying the Performance Trade-offs of ANNS on Milvus.
Authors: Turan, Melek1 (AUTHOR) melek.turan@ozdilek.com.tr, Peker, Musa1 (AUTHOR) musa.peker@ozdilek.com.tr, Kocakır, Nadir1 (AUTHOR) nadir.kocakir@ozdilek.com.tr
Source: International Journal of Software Engineering & Knowledge Engineering. 9/30/2026, Vol. 36 Issue 12, p1673-1699. 27p.
Subject Terms: *Distributed computing, Computer performance, Benchmark problems (Computer science)
Abstract: Efficient similarity search over high-dimensional embeddings is a core requirement of modern AI systems, yet most evaluations of Approximate Nearest Neighbor Search (ANNS) methods overlook the system-level dynamics that govern real-world performance. This study presents a comprehensive benchmark of six widely used ANNS algorithms — FLAT, IVF-FLAT, IVF-SQ8, HNSW, DiskANN, and ScaNN — integrated into Milvus and evaluated across both standalone and distributed (k3s-based) deployments. To capture modality-dependent behavior, the analysis spans two medium-size datasets, each consisting of 1.6 million embeddings: a large Turkish text corpus and a visual dataset generated using a unified 768-dimensional representation space. Each algorithm is assessed using recall@10, latency, throughput, and tail-latency (tp50–tp99), alongside detailed profiling of CPU utilization, RAM consumption, and disk I/O during indexing and querying. The results reveal distinct operational signatures: ScaNN and IVF-SQ8 achieve the highest throughput — particularly for image embeddings; HNSW offers superior recall but imposes greater memory pressure; and DiskANN provides stable, memory-efficient performance with a bounded recall ceiling for text. Distributed experiments further show that graph-based indexes benefit from multi-node parallelism, while disk-aware methods maintain consistent performance. Overall, the findings demonstrate that ANNS selection requires a holistic evaluation that accounts for both retrieval quality and system-level resource behavior under realistic workloads. [ABSTRACT FROM AUTHOR]
Copyright of International Journal of Software Engineering & Knowledge Engineering is the property of World Scientific Publishing Company and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database: Business Source Index
FullText Text:
  Availability: 0
Header DbId: bsx
DbLabel: Business Source Index
An: 193950458
RelevancyScore: 1435
AccessLevel: 6
PubType: Academic Journal
PubTypeId: academicJournal
PreciseRelevancyScore: 1435
IllustrationInfo
Items – Name: Title
  Label: Title
  Group: Ti
  Data: Search at Scale: Quantifying the Performance Trade-offs of ANNS on Milvus.
– Name: Author
  Label: Authors
  Group: Au
  Data: <searchLink fieldCode="AR" term="%22Turan%2C+Melek%22">Turan, Melek</searchLink><relatesTo>1</relatesTo> (AUTHOR)<i> melek.turan@ozdilek.com.tr</i><br /><searchLink fieldCode="AR" term="%22Peker%2C+Musa%22">Peker, Musa</searchLink><relatesTo>1</relatesTo> (AUTHOR)<i> musa.peker@ozdilek.com.tr</i><br /><searchLink fieldCode="AR" term="%22Kocakır%2C+Nadir%22">Kocakır, Nadir</searchLink><relatesTo>1</relatesTo> (AUTHOR)<i> nadir.kocakir@ozdilek.com.tr</i>
– Name: TitleSource
  Label: Source
  Group: Src
  Data: <searchLink fieldCode="JN" term="%22International+Journal+of+Software+Engineering+%26+Knowledge+Engineering%22">International Journal of Software Engineering & Knowledge Engineering</searchLink>. 9/30/2026, Vol. 36 Issue 12, p1673-1699. 27p.
– Name: Subject
  Label: Subject Terms
  Group: Su
  Data: *<searchLink fieldCode="DE" term="%22Distributed+computing%22">Distributed computing</searchLink><br /><searchLink fieldCode="DE" term="%22Computer+performance%22">Computer performance</searchLink><br /><searchLink fieldCode="DE" term="%22Benchmark+problems+%28Computer+science%29%22">Benchmark problems (Computer science)</searchLink>
– Name: Abstract
  Label: Abstract
  Group: Ab
  Data: Efficient similarity search over high-dimensional embeddings is a core requirement of modern AI systems, yet most evaluations of Approximate Nearest Neighbor Search (ANNS) methods overlook the system-level dynamics that govern real-world performance. This study presents a comprehensive benchmark of six widely used ANNS algorithms — FLAT, IVF-FLAT, IVF-SQ8, HNSW, DiskANN, and ScaNN — integrated into Milvus and evaluated across both standalone and distributed (k3s-based) deployments. To capture modality-dependent behavior, the analysis spans two medium-size datasets, each consisting of 1.6 million embeddings: a large Turkish text corpus and a visual dataset generated using a unified 768-dimensional representation space. Each algorithm is assessed using recall@10, latency, throughput, and tail-latency (tp50–tp99), alongside detailed profiling of CPU utilization, RAM consumption, and disk I/O during indexing and querying. The results reveal distinct operational signatures: ScaNN and IVF-SQ8 achieve the highest throughput — particularly for image embeddings; HNSW offers superior recall but imposes greater memory pressure; and DiskANN provides stable, memory-efficient performance with a bounded recall ceiling for text. Distributed experiments further show that graph-based indexes benefit from multi-node parallelism, while disk-aware methods maintain consistent performance. Overall, the findings demonstrate that ANNS selection requires a holistic evaluation that accounts for both retrieval quality and system-level resource behavior under realistic workloads. [ABSTRACT FROM AUTHOR]
– Name: AbstractSuppliedCopyright
  Label:
  Group: Ab
  Data: <i>Copyright of International Journal of Software Engineering & Knowledge Engineering is the property of World Scientific Publishing Company and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract.</i> (Copyright applies to all Abstracts.)
PLink https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=bsx&AN=193950458
RecordInfo BibRecord:
  BibEntity:
    Identifiers:
      – Type: doi
        Value: 10.1142/S0218194026500130
    Languages:
      – Code: eng
        Text: English
    PhysicalDescription:
      Pagination:
        PageCount: 27
        StartPage: 1673
    Subjects:
      – SubjectFull: Distributed computing
        Type: general
      – SubjectFull: Computer performance
        Type: general
      – SubjectFull: Benchmark problems (Computer science)
        Type: general
    Titles:
      – TitleFull: Search at Scale: Quantifying the Performance Trade-offs of ANNS on Milvus.
        Type: main
  BibRelationships:
    HasContributorRelationships:
      – PersonEntity:
          Name:
            NameFull: Turan, Melek
      – PersonEntity:
          Name:
            NameFull: Peker, Musa
      – PersonEntity:
          Name:
            NameFull: Kocakır, Nadir
    IsPartOfRelationships:
      – BibEntity:
          Dates:
            – D: 30
              M: 09
              Text: 9/30/2026
              Type: published
              Y: 2026
          Identifiers:
            – Type: issn-print
              Value: 02181940
          Numbering:
            – Type: volume
              Value: 36
            – Type: issue
              Value: 12
          Titles:
            – TitleFull: International Journal of Software Engineering & Knowledge Engineering
              Type: main
ResultId 1