• Welcome
  • Reviews
  • Videos
  • Gear
  • CyberGizmo Ethics Statement
  • Search
DJ Ware - The CyberGizmo
The Future is Now!
Skip to content
  • Welcome
  • Reviews
  • Videos
  • Gear
  • CyberGizmo Ethics Statement
  • Search
DJ Ware - The CyberGizmo

Videos

Home Videos
Videos

DJ Ware

Technology did not begin yesterday.

CyberGizmo explores Linux, security, networking, servers, network storage, and modern computing through the lens of history, experience, and real-world architecture. Many of today's "new" ideas are not new at all—they are old lessons rediscovered, forgotten, or repeated.

Through analysis, storytelling, humor, and practical experience, I examine what works, what doesn't, and what we can learn from the past before making the same mistakes again.

Welcome to CyberGizmo.

DJ Ware
Ransomware did not start with Bitcoin. It started with a floppy and a Panama PO box.

IIn this video I show you the one tool which beats ransomware.

Homework: one folder that would ruin your month, copied to a disk you unplug.

Next: living off the land — stealing the same box with tools that already shipped with Linux.

1989 AIDS/PC Cyborg Trojan (Joseph Popp)
This is a demonstration on an isolated VM. Do not run destructive commands on a machine you care about.
Most people still treat a firewall like a front door. Lock it down, keep the bad stuff out, and call it done.

That model is incomplete.

In this video I walk through why “build a firewall from scratch” is close but not the right question in 2026, and what actually matters instead:

- The dual role of a firewall (ingress + egress)
- Why outbound control belongs at the edge
- How to approach default-deny without unplugging yourself
- Why high-value hosts still need their own controls
- A practical sequence that won’t drive you crazy

I’m not going to hand you a 200-line ruleset. I’m going to show you the architecture that makes the ruleset worth writing.

**Chapters**
00:00 - Intro
00:51 - Decade 1990
01:00 - egress firewalls
01:17 - packet filter firewalls
01:51 - Stateless Firewalls
02:31 - egress filters
03:19 - Unified Threat Manament (UTM)
03:30 - Threats ni 2026
03:39 - PROMPTFLUX and PROMPTSTEEL
04:07 - Checkpoint Says
04:28 - NGFW
05:10 - Another Way
05:41 - egress
07:23 - Nexus Says my Computer isnt Genuine
09:13 - nftables + egress
09:32 - PiHole and pfBlockerNG
12:43 - Egress Monitor
I’ve been using Linux for three decades (In November 2026).

Is Linux  perfect?
Is Linux Easy?
...and the lessons I finally learned.

In this video I walk through eight of them:

• Why I run different distros on different machines
• How I decide how fast things change
• Why I learn the operation instead of the command
• When I use the GUI and when I drop to the shell
• How Linux once showed me a project was going to fail — and I listened
• How to make Linux fit the job
• And why, after all these years, I’d still choose it again

It’s the hard-won perspective that comes from actually using the thing for a long time.

Timestamps 
00:00 - Intro
00:49 - Linux Gives Me Choices
03:42 - Decide How Fast Things Change
04:46 - Learn the Operation
06:23 - GUI and Shell
08:57 - I Can Build THings in the Shell
11:00 - Linux Let's Me See Hwat Happenning
15:55 - I Can Make Linux Fit the Job
19:50 - Linux Keeeps Me Learning
I tested ext4, XFS, Btrfs, ZFS, and F2FS on the same hardware to answer a more useful question than simply, “Which filesystem is fastest?”

Instead of dragging you through 26 benchmark charts, I distilled the results into four real-world workloads:

Gaming • Workstation • Server • NAS

The testing includes small and medium workloads, single-user and multi-user behavior, and several different kinds of read and write activity.

But graphs are never the full story.

A filesystem may win a throughput benchmark while consuming more memory or CPU, behaving differently as the workload grows, or introducing operational tradeoffs that matter more than a few percent of performance. Your workload may also be very different from mine.

The workload model used in this video is new, and it needs more data points before I’m comfortable treating it as anything more than a useful decision-making tool.

So this isn’t “here’s the winner.”

It’s what do you gain, what do you give up, and which filesystem makes sense for what you actually do?

If you want to see the raw benchmark charts and the work behind the scoring model, let me know in the comments. I may do a behind-the-scenes or live session and go through the data.

CHAPTERS
00:00 - Intro
00:24 - Methodology
00:54 - EXT Filesystem
02:12 - XFS
02:47 - BtrFS
02:58 - ZFS
03:58 - Test Machine
05:36 - WD SN850X Specs
05:59 - Which one Won?
06:07 - Use Cases
06:27 - Gaming
06:50 - Workstatoion
07:14 - Servers
07:33 - NAS
07:50 - Small File Size Performance
09:53 - Medium File Size Performance
11:37 - Wrapup
This Mini PC Idles at 4W… Then Hits 90°C (GMKtec M3 Pro + Aurora Linux Server Test)

Today I’m putting the GMKtec NucBox M3 Pro (32 GB DDR4-3200 + 1 TB NVMe) through a real server-oriented test using Aurora Linux — an immutable Fedora Atomic-based system.

GMKtec Official Website:
https://tinyurl.com/DJWarre-GMKtec-M3Pro

Amazon:
M3 Pro 16+512:
https://www.amazon.com/dp/B0C3HKXNKD?maas=maas_adg_AD875FB724D0CBA4A78526984D3B5F17_afap_abs&ref_=aa_maas&tag=maas

M3 Pro 32+1:
https://www.amazon.com/dp/B0FLJSB6KC?maas=maas_adg_4CD62DC5193D5A71ACE6549FE5A73069_afap_abs&ref_=aa_maas&tag=maas&th=1

50$ OFF Code: M3PROKOL

We cover:
• Full Aurora Linux install as a headless server
• Phoronix benchmarks
• Detailed thermal testing with s-tui + stress-ng
• Real power draw numbers (idle & under load)
• Cooling behavior and recovery times
• A short post-mortem on the Void Linux + KDE attempt

Key numbers from this unit:
• Idle power: 3.3 – 4.4 W
• Full stress: 40 – 45 W
• Peak temperatures under load: 86–91°C (Thermal Max 100°C)
•  Metal case helps dissipates heat and aids in cooling
• Cool-down behavior:
   Distrobox: 4 minutes
   rpm-ostree: 2 minutes

Timestamps:
00:00 - Intro
00:18 - Testing GMKtec Nucbox M3 Pro and Audria Linux
00:42 - M3 Pro SPecs
00:55 - Test Methodology
01:11 - Specs
01:38 - GPU
01:43 - Expansion
02:10 - Portd
02:36 - Networking
02:58 - Dimensions
03:10 - Cooling
03:14 - Who is this for?
03:33 - Aurora Linux
06:33 - Benchmarks
08:08 - Summary of Benchmarks
08:58 - Conclusions
10:26 - Pros and Cons

Links:
• Aurora Linux → https://getaurora.dev

If you found this useful, drop a like and let me know in the comments what mini PC you’re currently running as a server or always-on node.

#GMKtec #M3Pro #MiniPC #AuroraLinux #Homelab #Linux #Server
In 2018 I recorded a video about Fedora Silverblue 29… but I never released it. I was completely lost.

The first Silverblue video I actually published was on Fedora 30 in April 2019. A comment on the Fedora 31 video later helped a little: “This looks like a version control system for the distro.” Partially true — but atomic is more than that. It’s a separation of the filesystems: the parts that can be read-only and the parts that need read/write.

It took until Silverblue 34 before I trusted it as a daily driver. Later I tried Kinoite, bounced back because of hardware, fought with a hot Intel laptop and an AtomMan X7 Ti, and finally landed on proper AMD hardware in 2025. In April 2026 I installed Bluefin while keeping Kinoite on the second NVMe.

This is the full story — the wins, the hardware failures, the dual-drive experiment, and why I haven’t switched back to Kinoite since mid-June 2026.

We also cover:
• Official Atomic (Silverblue/Kinoite) vs curated Universal Blue (Bluefin/Aurora)
• Toolbox vs Distrobox
• Update cadence differences
• How clean I keep the rpm-ostree base
• Why Universal Blue recommends clean installs when switching between GNOME and KDE images

Chapters below.

If you’re deciding between pure Fedora Atomic and Bluefin/Aurora, this should help.

Links:
• Project Bluefin: https://projectbluefin.io
• Aurora: https://getaurora.dev
• Universal Blue: https://universal-blue.org
• Fedora Atomic Desktops: https://fedoraproject.org/atomic-desktops/

Chapters below.
00:00 - Intro
01:37 - Fedora 30
01:58 - Fedora 30
02:44 - SilverBlue34
04:29 - Kinoite 35
07:41 - GMKtec EVO-X1 and Kinoite 44
08:53 - Bazzite
12:39 - How do i use this??
14:32 - Final Thoughts
Zero Persistence Prototype – First real test on the Icewhale ZimaBoard 2 ($339, Intel N150, 8 GB RAM).

In this video I take the ideas from “The OS is Dying” and put them into practice on actual hardware. We build a working Zero Persistence environment using Firecracker microVMs on Ubuntu 24.04 and see how this low-power board performs under real workloads.

What you’ll see:
• Full ZimaBoard 2 specs & first impressions
• Why Zero Persistence matters
• Firecracker microVM setup
• Ephemeral demo (files should disappear after reboot)
• Power, thermals & real-world performance
• Benchmark: ZimaBoard 2 (N150) vs Brand-X (N97)
• Who this board is best for

This is Part 1 of the Zero Persistence series. Part 2 will push multi-VM density, kernels, and deeper testing.

Timestamps:
00:00 - intro
00:24 - So many comments
00:40 - Icewhale Zimaboard 2
01:39 - WHat's in the Box?
02:57 - First Impressions
03:17 - Marker 6
06:20 - Zero Persistence Demo
14:54 - Benchmark N150 vs N97
15:52 - Final Thoughts

Hardware tested:
• Icewhale ZimaBoard 2 (Intel N150, 8 GB LPDDR5)
• Brand-X N97 system (for comparison)

Links:
• Official website link: https://bit.ly/4eav6l1 
• Amazon US:https://amzn.to/48qzrgF
• Firecracker docs → https://firecracker-microvm.github.io/
• Previous video: The OS is Dying → https://www.youtube.com/watch?v=YFk5hodJqxc&t=11s

What should we test next? Drop your ideas in the comments.

#NAS #ZimaBoard #Server
We set out to make a simple video about hardening logins. Instead, we discovered how deeply broken our cloud identity model has become.

From passkeys that quietly bypass your second factor, to “Sign in with Google” giving providers authority across dozens of sites, the convenience we’ve embraced has created massive single points of failure.

In this video I walk through real account takeover cases, the limitations of current passkey revocation, and a practical compartmentalization strategy to limit blast radius when (not if) something goes wrong.

Links:
NIST Digital ID Guidelines: https://nvlpubs.nist.gov/nistpubs/SpecialPublications/NIST.SP.800-63b-4.pdf
Trusted Security Group: TCG Storage Security
Subsystem Class (SSC):
Opal - https://trustedcomputinggroup.org/wp-content/uploads/TCG-Storage-Opal-SSC-v2.30_pub.pdf
FIDO Alliance : Trusted AI Agents - https://fidoalliance.org/fido-alliance-to-develop-standards-for-trusted-ai-agent-interactions/

Chapters below 
00:00 - Intro
00:22 - Problem
01:53 - 2FA
02:30 - MFA
02:51 - passkey
04:22 - What is 2FA Suppose to Mean?
05:01 - Revoking a passkey
05:20 - Login in AS ...
06:14 - The Cloud Trap
07:20 - Best Practices
07:37 - Simple Best Practices
08:10 - Oh and one more thing
10:54 - Closing
In 2026, AI is accelerating zero-day discovery and stealth malware while the old "build and maintain" model for our systems is breaking down. Corporate security's "find and fix" approach can't keep up — once malware gains a foothold, you may have only minutes before it spreads.
DJ explores why persistent systems have become major liabilities and proposes a fundamental shift: Zero Persistence. Learn how ephemeral systems, on-demand services, and moving-target designs can help us reclaim control.

00:00 - Intro
02:38 - Why Does the Old Model Fail?
04:13 - Docker Best Practices
05:39 - A Proposal and Open Discussion
07:31 - Fundamental Shift in Thinking about Systems
08:07 - Call to Action
Subscribe

Recent Posts

  • KDE neon
  • Ghost Browser Introduction
  • Linux Internals: Storage
  • Red Hat Enterprise Linux 9
  • Asahi Linux – First Look

Recent Comments

No comments to show.

Archives

  • September 2022
  • August 2022
  • July 2022

Categories

  • Blog
  • Initial Look
  • Reviews
  • Uncategorized
Back to Top
©2022 DJ Ware - The CyberGizmo
Powered by Anima & WordPress.