StorageReview.com
@storagereview

Enterprise infrastructure since 1998. Storage, servers, networking, AI. Independent reviews, lab benchmarks, breaking news from the data center.

 Cincinnati, OH
storagereview.com/storage_ne…
Joined November 2011
5,423 Photos and videos
One Micron 245TB SSD replaces eight spindles and hands back 55W per server. Swap 17,456 HDDs for 2,182 of them and the freed 120kW runs an entire GB200 NVL72. An exabyte fits in 6 racks instead of 22. storagereview.com/review/mic… #DataCenter #AI @MicronTech

Micron 6600 ION 245TB: Swap the Hard Drives, Power an NVL72 for Free

Micron 6600 ION 245TB measured: one SSD replaces eight HDDs, cuts 55W per unit, and puts an exabyte in 6 racks instead of 22.

storagereview.com
3
13
94
46,466
NetApp is acquiring PEAK:AIO. We clocked PEAK:AIO at 160 GB/s over NVMe-oF RDMA from one Dell R7725xd; now its scale-out pNFS metadata work heads toward ONTAP. Curious to see where NetApp takes it. storagereview.com/news/netap… #AIStorage #pNFS @NetApp

NetApp to Acquire PEAK:AIO, Bringing Its Scale-Out pNFS Metadata Work to ONTAP

NetApp plans to acquire PEAK:AIO and pair its scale-out pNFS metadata work with ONTAP for AI clouds targeting trillions of files.

storagereview.com
241
Supermicro is now shipping NVIDIA Vera Rubin NVL72 racks with DLC-2 liquid cooling and 1.8MW in-row CDUs. One Scalable Unit: 16 racks, 1,152 Rubin GPUs, 331 TB of HBM4. storagereview.com/news/super… #VeraRubin #LiquidCooling #AIInfrastructure @Supermicro @NVIDIAAI

Supermicro NVIDIA Vera Rubin NVL72 Racks Now Shipping With 1.8MW In-Row CDUs and a 1,152-GPU...

Supermicro NVIDIA Vera Rubin NVL72 racks now ship with DLC-2 cooling, 1.8MW in-row CDUs, and a 1,152-GPU, 16-rack Scalable Unit blueprint.

storagereview.com
1
1
2
297
The RTX PRO 4500 Server Edition delivered a median 2.7x the L4's BF16 output throughput in the same HPE DL145 Gen11 edge server, with sub-second TTFT where the L4 queued, for only 105W more. storagereview.com/review/nvi… #EdgeAI #vLLM #Blackwell @HPE_Compute @NVIDIAAI

NVIDIA RTX PRO 4500 Edge Review: 2.7x the L4 Inside an HPE ProLiant DL145 Gen11

RTX PRO 4500 edge review: the Blackwell Server Edition vs L4 in an HPE ProLiant DL145 Gen11, 2.7x the tokens, sub-second TTFT, 105W more.

storagereview.com
2
4
490
Inside the ASUS ExpertCenter Pro ET900N G3 GB300 DGX Station: fan bracket out to trace the loop, cold plates on the Superchip, SOCAMM, and ConnectX-8, a small fan for the 400G optics, and carry handles on top. Review on the site now. #ASUS #GB300 #DGXStation @ASUS @ASUSUSA
2
3
488
Giga Computing's GAIFA is live in Taiwan: six liquid-cooled racks at a planned 700 kW, two of them NVIDIA GB300 NVL72, for validating AI infrastructure with customers before it ships. storagereview.com/news/giga-… #AIFactory #GIGABYTE #DLC @GigaComputing @NVIDIAAI

Giga Computing Opens GAIFA, a 700 kW Liquid-Cooled AI Factory Test Bed With Two GB300 NVL72 Racks

Giga Computing GAIFA: a 700 kW, six-rack, direct-liquid-cooled AI factory test bed in Taiwan with two NVIDIA GB300 NVL72 racks.

storagereview.com
2
3
10
582
VAST DataEnclave puts AI models and regulated data inside NVIDIA Confidential Computing: encrypted guest memory, GPU memory, and NVLink, keys released only after attestation. Ships Q1 2027. storagereview.com/news/vast-… #ConfidentialAI #VAST #AI @VAST_Data

VAST DataEnclave Runs AI Models on Regulated Data Inside NVIDIA Confidential Computing, Ships Q1...

VAST DataEnclave brings NVIDIA Confidential Computing to the DataEngine: attested enclaves for models and regulated data, shipping Q1 2027.

storagereview.com
1
1
4
496
Fibre Channel's new FS-SP-3 security standard drops 3DES, MD5, and SHA-1, adds ML-KEM-1024 and ML-DSA-87 for CNSA 2.0, and trims the spec from 288 to 152 pages. Deep dive on what's new. storagereview.com/review/fci… #FibreChannel #PostQuantum #FCSP3

FCIA's FC-SP-3: A Standard Ready for the Quantum-Computing World

In enterprise IT, most security upgrades typically occur in response to breaches or other significant incidents. The recently developed INCITS FC-SP-3 standard, however, represents a proactive...

storagereview.com
3
509
IDC: external storage hit $10.3B in Q2 2026, up 33.6%. All-flash took 52% of revenue, high-end grew 90.6%. Servers set a $166.3B record, up 52%, GPU server ASPs up 43.6% to ~$170K. storagereview.com/news/idc-e… #IDC #EnterpriseStorage #AIInfrastructure

IDC External Storage Tracker: $10.3 Billion in Q2 2026 as Server Market Tops $166 Billion on AI...

IDC external storage tracker: Q2 2026 revenue hit $10.3B, up 33.6%, as servers set a $166.3B record. Dell leads both; AFAs take 52%.

storagereview.com
3
495
ASUS Ascent QN10 review: Snapdragon X2 Elite, 18 Oryon cores, 80 TOPS NPU in a 0.7-liter mini PC. Geekbench 7 multi-core 24,049, 73% past the NUC 14 Pro, and it stays near silent. storagereview.com/review/asu… #ASUS #Snapdragon #MiniPC @ASUS @ASUSUSA

ASUS Ascent QN10 Review: 80 TOPS and 18 Oryon Cores in a 0.7-Liter Mini PC

We benchmark the ASUS Ascent QN10, the first Snapdragon X2 Elite mini PC: 18 Oryon cores, 80 TOPS NPU, and how it stacks up.

storagereview.com
4
682
QNAP TS-h966TX: 5 SATA bays plus 4 U.2 NVMe bays in a tower 36% smaller than its old 6-bay Thunderbolt NAS. Dual TB4, 10GbE, 2.5GbE, 6-core Core i3, QuTS hero ZFS. $1,299 on pre-order. storagereview.com/news/qnap-… #QNAP #NAS #Thunderbolt4 @QNAP_nas

QNAP TS-h966TX Combines Thunderbolt 4, 10GbE, and U.2 NVMe in a Compact 9-Bay NAS

QNAP TS-h966TX is a 9-bay Thunderbolt 4 NAS with five SATA bays, four U.2 NVMe bays, 10GbE and 2.5GbE, a 6-core Core i3, and QuTS hero ZFS.

storagereview.com
2
562
HPE Alletra Storage MP X10000 Release 4 is GA: 16 nodes and 16 JBOFs per cluster, ~23PB raw, roughly 2x the prior release. Native NFS, object, RDMA extends to file, KMIP and TLS 1.3 for air-gapped sites. storagereview.com/news/hpe-a… #HPE #ObjectStorage #EnterpriseStorage @HPE_Storage

HPE Alletra Storage MP X10000 Release 4 Is GA, Doubling to 16 Nodes and 23PB Raw and Adding Native...

HPE Alletra Storage MP X10000 Release 4 is GA: 16 nodes and 16 JBOFs, 23PB raw, native NFS beside object, RDMA for file, KMIP and TLS 1.3.

storagereview.com
2
4
578
The TrueNAS Proxmox Plugin turns each VM disk request into an OpenZFS zvol over iSCSI or NVMe/TCP, with native snapshots that capture RAM state. TrueNAS 25.10+, on GitHub now. storagereview.com/news/truen… #TrueNAS #Proxmox #OpenZFS @TrueNAS

TrueNAS Proxmox Plugin Turns Every VM Disk Request Into an Automated zvol Over iSCSI or NVMe/TCP

TrueNAS Proxmox Plugin provisions VM disks as OpenZFS zvols over iSCSI or NVMe/TCP, with native snapshots and no manual LUN work.

storagereview.com
1
1
9
611
CoreWeave brought up multi-rack Vera Rubin NVL72: 72 GPUs per Dell-built rack, 1.6 Tb/s per GPU, plus a no-fee Archive tier for AI Object Storage. storagereview.com/news/corew… #VeraRubin #CoreWeave #AIInfrastructure @CoreWeave @Dell @NVIDIAAI

CoreWeave Deploys Multi-Rack Vera Rubin NVL72 Cluster: Hundreds of Rubin GPUs, 1.6 Tb/s Per GPU,...

CoreWeave's multi-rack Vera Rubin NVL72 cluster links hundreds of Rubin GPUs at 1.6 Tb/s, with cross-region writes and an Archive tier.

storagereview.com
1
7
561
192GB of unified memory, 160GB of it for the GPU. The HP ZBook Ultra G3a 16 has the most GPU memory we've seen on a Ryzen AI Max system, plus a 100W thermal design, up from 55W. Preview and teardown: storagereview.com/review/hp-… #HPZBook #LocalAI #Workstation @ZbyHP

HP ZBook Ultra G3a 16 Preview: 192GB of Unified Memory Aims for the Top of the Local AI Laptop...

HP ZBook Ultra G3a 16 preview: 192GB unified memory, 160GB for the GPU, and a 100W thermal redesign. Benchmarks when it ships in October.

storagereview.com
1
4
37
4,251
One 512GB DDR5 RDIMM at 16W vs. four 128GB modules at 44.2W. Micron's TSV-stacked module puts 12TB in a dual-socket server at up to 9,200 MT/s. AMD and Intel validating; volume 2H 2027. storagereview.com/news/micro… #Micron #DDR5 #DataCenter @MicronTech

Micron Shows off 512GB DDR5 RDIMM: 12TB per Dual-Socket Server at 9,200 MT/s, Volume Production in...

Micron's 512GB DDR5 RDIMM: 12TB in a dual-socket server at up to 9,200 MT/s and 16W per module. AMD and Intel validating; volume 2H 2027.

storagereview.com
3
18
897
192GB of unified memory, up to 160GB of it for the GPU: the HP ZBook Ultra G3a 16 has the most GPU memory we've seen on an AMD Ryzen AI Max system. New 100W thermal design (55W on the G1a), 96Wh battery, Gen5 slot to 8TB, 4.2 lb. #HPZBook #LocalAI #Workstation @ZbyHP @AMDRyzen
30
71
714
66,271
Seagate and WD AI storage studies: storage infrastructure (43%) outranks compute (27%) as the top AI deployment challenge, and 74% of enterprises are keeping data longer. storagereview.com/news/seaga… #AIStorage #HDD #DataCenter

Seagate and WD AI Storage Research Finds Enterprises Rank Storage Above Compute as the AI Bottleneck

Seagate and WD AI storage research: enterprises rank storage above compute as the AI bottleneck, and 74% are keeping data longer.

storagereview.com
1
5
550
MSI's XpertStation WS300 didn't throttle in our review, now here's the data: 75 min burn-in at ~1,000W GPU, Blackwell Ultra die held at 60C, clock flat at 2.07GHz, coolant 10C rise across the loop. storagereview.com/review/msi… #MSI #DGXStation #GB300 @msitweets @NVIDIAAI

MSI XpertStation WS300 Thermals: Why a 1,300W GB300 Doesn't Throttle on a Desk

MSI XpertStation WS300 thermals: MSI cold plate and radiator data plus our GB300 run show Blackwell Ultra holding 2.07GHz, no throttling.

storagereview.com
3
9
621
Lightbits Inferra virtualizes GPU memory across DRAM and NVMe, keeping the KV cache alive across tiers. Lightbits claims 16x more concurrent sessions, >100x lower latency vs recompute, and 10M-token contexts. storagereview.com/news/light… #KVCache #AIInference #NVMe @LightbitsLabs @solidigm
1
1
557
Load more