On-Demand Presentation

How to use HCI and Hard Drives to Fight Data Center Inflation

Watch a running VM move from flash to hard drives and back without going offline, and see how VergeOS runs most workloads on hard drives at a fraction of the cost of flash.

GC
George Crump
George Crump
Founder, Storage Switzerland
DV
Dave Vincent
Dave Vincent
Technical Solutions Strategist
On-Demand · Includes a Recorded Tier-Move Demo
TIER 1 · NVMe FLASH Live move VM stays online TIER 4 · HARD DRIVES Workload VM online
01 / 13
Today’s Agenda

Six Steps to Cheaper Capacity

01
Inflation
Data Center Inflation Reaches the Drive Bay
Flash and memory prices climbed 70 to 95 percent in the first half of 2026, and AI demand keeps the pressure on.
02
Economics
All-Flash by Default Is a Budget Decision
Every HCI platform ships all-flash, so every workload pays flash prices whether it needs the performance or not.
03
Control
Pick the Tier and Change Your Mind
Move running VMs between flash and hard drives on command, in either direction, with no downtime.
04
Recorded Demo
A Live SQL Disk Moves to Hard Drives
A SQL Server data disk moves from NVMe to hard drives while the VM keeps serving queries.
05
Performance
How VergeOS Makes Hard Drives Perform
Read prefetching, write reordering, RAM caching next to the VM and global inline deduplication.
06
Protect
Protect High-Capacity Drives
Parallel rebuilds, frequent snapshots and ioGuardian keep the data on large drives protected.
verge.io
02 / 13
01 · Inflation

Data Center Inflation Reaches the Drive Bay

STREET PRICE PER TERABYTE Solidigm D7-PS1010 · 15.36 TB NVMe $814 Kioxia KCMYXRUG15T3 · 15.36 TB NVMe $706 WD Ultrastar DC HC580 · 24 TB hard drive $56 Flash costs 12 to 14 times more per terabyte
ServerSupply street prices checked September 28, 2026. Prices move week to week.
70–95%
Flash and memory price climb
First half of 2026
12–14×
Flash premium per terabyte
15.36 TB NVMe vs. a 24 TB hard drive
8
Hard drives for one SSD
One SSD’s price buys 192 TB of raw hard drive capacity
AI
Keeps the pressure on
AI buildouts absorb new fab capacity as fast as it comes online
Plan your next refresh as if this pricing is permanent.
verge.io
03 / 13
02 · Economics

All-Flash by Default Is a Budget Decision

  • The platform made the choice. HCI vendors assumed flash would keep getting cheaper, so all-flash became the default.
  • Every terabyte pays twice. Refresh capacity pays the flash price, and capacity-based licensing charges for it again.
  • Most workloads sit idle. File shares, VDI profiles, dev and test, log retention and general-purpose VMs rarely need flash.
  • Databases are the exception. Random-I/O workloads stay on flash. Everything else deserves a choice.
All-flash default: every VM pays flash prices Placed by need: flash for the working set only Flash Hard drive
Illustrative mix. Most environments need flash for far fewer VMs than they run on it today.
Hard drives are slower than flash. They are not tape.
verge.io
04 / 13
02 · Economics · Recorded Demo

See Where the Capacity Lives on VergeOS

  • One cluster, one vSAN. The dashboard shows four storage tiers online in the same VergeOS cluster.
  • Flash and hard drives side by side. Tier 1 is NVMe. Tiers 4 and 5 are hard drives.
  • Every tier at a glance. Used, used inflated, deduplication ratio and capacity for each tier.
  • Deduplication everywhere. It runs inline on every tier, flash and hard drive alike.
verge.io
05 / 13
03 · Control

Pick the Tier and Change Your Mind

How a tier move works (VergeOS) SQL Server VM Online · serving queries OS 80 GB Tier 1 Data 200 GB Tier 1 moving… Tier 4 Log 40 GB Tier 1 TempDB 40 GB Tier 1 Backup 150 GB Tier 4 Storage media cost $263 Saved $130 · 49% Tier 1 · NVMe Performance: OS, transaction log, tempdb Tier 4 · HDD High capacity, cost-effective OS80 GB Log40 GB TempDB40 GB Backup150 GB Data200 GB Data200 GB Preferred Tier changed: Tier 1 → Tier 4 · live, online background migration Move complete · no downtime Capacity after the move Flash 160 GB HDD 350 GB · 69% of the VM
  • Pick the tier. Change one setting on the drive and VergeOS moves the data in the background.
  • Change your mind. Move it back to flash the same way. The VM stays online, with no maintenance window.
  • 1.3 seconds. API acknowledgment in VergeIO lab testing. A 25 GB disk finished moving in 13 seconds.
  • $263 to $133. The VM’s storage media cost drops 49%.
Cost basis: raw media at $706/TB for a 15.36 TB NVMe SSD and $56/TB for a 24 TB HDD, street prices September 28, 2026. Excludes redundancy copies.
verge.io
06 / 13
04 · Recorded Demo

A Live SQL Server Disk, Flash to Hard Drives and Back

  • Two screens. VergeOS on the left. On the right, a monitor querying a live SQL Server VM every second.
  • Out and back. The SQLDATA disk moves from Tier 1 NVMe to Tier 5 hard drives, runs there, then returns to Tier 1.
  • No interruption. Rows per second and availability hold through both moves. The VM is never migrated or restarted.
  • Flash freed for the working set. Bulk data lands on hard drives at $56 per TB instead of $706 per TB on NVMe.
verge.io
07 / 13
04 · Recorded Demo · AI Agent

Let an AI Agent Find the Long Tail

  • Two screens. VergeOS on the left, with demo-archive-1’s disk on Tier 1 NVMe. An AI agent on the right.
  • Guardrails first. The agent runs on a model on the lab’s own GPUs, sees seven demo VMs, and can change only a drive’s preferred tier.
  • Data, not guesses. Asked which VMs are idle, it reads a report built from VergeOS drive statistics.
  • Tail to disk, busy VMs on flash. It moves the four idle archive VMs to Tier 5 hard drives and leaves the three busy VMs, about 130 to 150 IOPS each, on flash.
verge.io
08 / 13
05 · Performance

How VergeOS Makes Hard Drives Perform

VM RAM CACHE SERVER 1 SERVER 2 SERVER 3 1 2 3 4 5 6 SEQUENTIAL READ · PREFETCH IN PARALLEL THE VM READS FROM RAM
01Read Prefetching
Sequential reads are pulled in parallel from drives across the cluster into RAM.
INCOMING RANDOM WRITES WRITE BUFFER 4 1 6 2 5 3 1 2 3 4 5 6 123456 RANDOM WRITES ARRIVE REORDERED AND GROUPED IN THE BUFFER ONE SEQUENTIAL RUN · NO WASTED SPIN
02Write Reordering
Random writes become sequential runs before they reach the platter.
VM 1 VM 2 VM 3 STORED ONCE A B C D E F G A B C A D B A C E A F C B G A D E A A B C A D B A C E A F C B G A D E A 18 BLOCKS WRITTEN BY THREE VMS 7 STORED · DUPLICATES REFERENCED
03Global Inline Deduplication
Duplicate blocks are stored once, so more of the working set fits in cache.
VergeOS runs separate code built for spinning disk. The guest does nothing to earn it.
verge.io
09 / 13
06 · Protect

Protect High-Capacity Hard Drives

SERVER 1 SERVER 2 SERVER 3 24 TB DRIVE FAILED REBUILDING FROM EVERY DRIVE AT ONCE REDUNDANCY RESTORED
01Parallel Rebuilds
Every drive in the cluster helps rebuild a failed one.
VMHEALTHYBAD CHANGERESTORED -6h -5h -4h -3h -2h -1h HOURLY SNAPSHOTS
02Frequent Snapshots
Hourly recovery points for every VM on the hard drive tier.
THIS SITEREMOTE VERGEOS BLOCKS LOST BEYOND REDUNDANCYSELF-HEALED FROM REMOTE COPY
03ioGuardian
Blocks lost beyond redundancy stream back from a remote site.
Large drives cut the cost per terabyte. The platform keeps the data on them protected.
verge.io
10 / 13
What Makes Us Different

Every Function. Unified Infrastructure.

Hypervisor Storage Networking Backup Disaster Recovery API API API API API VergeOS Infrastructure OS Unified infrastructure Hypervisor Storage Networking Backup DR
Five products, five upgrades, five vendors. One platform, one unified infrastructure.
  • VergeOS runs compute, storage, networking, and data protection as one unified infrastructure.
  • One install turns standard x86 servers into a complete infrastructure, including the DR site.
  • One upgrade moves the whole environment forward, so the maintenance window shrinks to hours.
  • An operating system, not a stack, means less overhead and a smaller footprint than assembled products, so the same physical servers hold more VMs.
verge.io
11 / 13
Why You Should Care

Simpler to Run, Cheaper to Own, Ready for What Is Next

IT leaders are asked to run more sites, more workloads, and more data with the same team. The platform has to give that time back.

01
Simplify Your Life
One interface covers virtual machines, storage, networking, snapshots, and failover. Your team stops moving between four consoles, four support contracts, and four upgrade calendars.
02
Lower Cost
A single license replaces the hypervisor, the array software, and the backup product. The servers stay. The savings repeat at every renewal rather than once at purchase.
03
Future Ready
Add a second site, a new tenant, or a GPU cluster for AI from the same platform. The architecture holds as the requirements change, which protects the decision you make today.
verge.io
12 / 13
Thank You for Watching

Buy Flash for the Working Set and Hard Drives for the Rest

VergeOS runs flash and hard drives in one HCI cluster, moves running VMs between them on command, and licenses per server, so the savings stay in your budget.

Schedule a Technical
Virtual Whiteboard Session
meetings.hubspot.com/gcrump/technical-whiteboard-session
George CrumpFounder, Storage SwitzerlandDave VincentVergeIO
13 / 13
Arrow keys or Previous and Next to advance