Publications

Technical analysis of storage technologies

• VAST Data's '29x Data Reduction' Claims: The Storage Industry's Most Brazen Lies

  VAST Data's co-founder claims customers see 8x to 29x capacity advantages versus HDFS. We expose the fraudulent math, the deliberate straw man, and why VAST has become the storage industry's most prolific source of misleading claims.

• Pure Storage's FlashBlade//EXA '10 TB/sec' Claim: When Vague Numbers Replace Real Benchmarks

  Pure Storage claims FlashBlade//EXA delivers 'more than 10 TB/sec' read performance. We analyze why this vague claim tells us almost nothing about real-world performance.

• VAST Data's 99.9991% Uptime and 10x Kafka Claims: The New Standard for Unverifiable Marketing

  VAST Data claims 99.9991% measured uptime (4.7 minutes downtime per year) and 10x performance advantage over Kafka. We analyze what these numbers actually mean and why they're nearly impossible to verify.

• DDN's '11x Faster' IO500 Claims: What the Benchmark Actually Measures

  DDN claims it delivers '11x more AI training runs per day' based on IO500 benchmark results. The claim confuses benchmark score with operational capability, excludes higher-scoring competitors, and uses significantly different hardware configurations.

• Hammerspace's 'Standard NFS' Achievement: A Technical Reality Check

  Hammerspace achieved #18 on IO500 using pNFS v4.2, claiming this proves 'standard protocols' can deliver HPC-class performance. The technology is legitimate. The positioning about 'standard' is misleading.

• HPE Alletra's DASE Strategy: Learning From VAST Without the IP Risk

  HPE is adopting VAST's disaggregated architecture concept for Alletra MP but implementing it two ways: partnering with VAST for file storage, developing independently for block storage. The strategy avoids litigation but creates different operational risks.

• IBM's $11B Confluent Acquisition: Event Streaming Infrastructure, Not an AI Platform

  IBM acquired Confluent for $11 billion to create a 'Smart Data Platform for Enterprise Generative AI.' The technical reality: Confluent provides Kafka-based event streaming. That solves specific problems well. AI is not one of them.

• VAST Data's $30B Valuation Ignores Operational Complexity Costs

  VAST claims 90% gross margins enabling $30B valuation. But VAST's three-tier protection architecture, proprietary algorithms, and metadata asymmetries create operational complexity that will compress margins through support costs. Investors are underestimating this risk.

• MinIO ExaPOD: Credible Architecture, Questions on Methodology

  MinIO's exabyte-scale reference architecture makes defensible claims backed by historical benchmarks. We examine what's verifiable, what needs more transparency, and how the commodity approach compares to appliance vendors.

• Pure Storage FlashBlade//EXA: Verified Benchmarks vs. Marketing Claims

  Pure Storage submits to STAC-M3 auditing—a credible approach. But claims of '30% better than competitors' and '10 TB/s from early testing' deserve the same scrutiny applied to any vendor.

• VAST DataBase Benchmarks: The Numbers We Can't Verify

  VAST Data claims 25% faster than Iceberg, 60x faster updates, and 500M messages per second. The numbers are impressive—but without published methodology or independent verification, they remain marketing claims, not engineering data.

• Weka's SPECstorage Records: How Benchmark Transparency Should Work

  Weka set #1 rankings across all five SPECstorage 2020 workloads in January 2025. More importantly, the results are independently audited and publicly verifiable—the standard every vendor should meet.

• When 53% of Vendors Are 'Leaders': The GigaOm Primary Storage Radar and Analyst Report Theater

  Analysis of GigaOm's Primary Storage Radar methodology - where 10 out of 19 vendors achieve 'Leader' status, 'Outperformer' means predicted future development, and customers need to understand what radar positioning actually tells them.

• NetApp's Disaggregated ONTAP and AI Data Engine: Marketing Meets Architecture

  Critical analysis of NetApp's AFX announcement - from the '1 exabyte effective capacity' claim to AIDE's four bundled solutions looking for problems, and why 'proven over decades' doesn't apply to new architectures.

• Scality's 'Pipelines Over Models' Argument: When Storage Vendors Discover AI

  Scality's CTO argues that data pipelines matter more than AI models. Convenient positioning for an object storage vendor - and partially true, but the two-tier storage simplification obscures real infrastructure complexity.

• VAST Data's 'Classical HPC' Framing: When Marketing Rewrites Storage History

  VAST Data positions legacy parallel file systems as obsolete for AI. The technical claims deserve scrutiny - metadata matters, but the 'classical vs. modern' narrative obscures more than it reveals.

• Cloudian's '26 Nines' Durability Claim: When Marketing Exceeds the Age of the Universe

  Mathematical analysis of Cloudian HyperStore's durability claims, from the defensible 14 nines to the physically impossible 26 nines - and why vendors publish numbers that dwarf the universe's existence.

• Dell ECS 'Eleven 9s' Durability: The Claim Without the Calculation

  Dell ECS and ObjectScale claim 99.999999999% durability using standard 12+4 Reed-Solomon erasure coding. The math should be straightforward. So why won't they publish the calculation?

• The Benchmark Problem: When Storage Vendors Claim 'Record-Setting' Performance Without Showing the Tests

  Analysis of unverifiable performance claims from Cloudian and Dell ObjectScale - from '74% improvement' to 'world's fastest' - and why benchmark numbers without methodology are marketing, not engineering.

• The VAST Data Marketing Machine: When Tech Journalism Becomes Promotional Content

  A critical examination of unverifiable claims, promotional language, and the troubling pattern of tech media publishing vendor narratives without scrutiny.

• Understanding VAST Data's Erasure Coding Architecture

  Technical analysis of VAST's three-tier protection architecture, examining metadata write asymmetry, operational complexity, and trade-offs versus traditional Reed-Solomon systems.