myvirtualcloud.net

Tools

Sat, 19 Sep 2026 16:00:00 +1200

Things I built because I needed them. Each one has its project on GitHub.

Ollama GPU Calculator

Check whether a GPU can run an Ollama model, and see estimated VRAM, tokens per second, and power. Pick a model size and a card; the readout updates as you go.

Bio

Fri, 18 Sep 2026 08:00:00 +1200

I’m Andre Leibovici. I started this site in 2009 as a place to write down what I was actually running — virtual desktops, storage, hyperconverged systems, later Kubernetes and GPUs — before the slides caught up.

The work has moved around. São Paulo, Sydney, Silicon Valley, Auckland. Startups I founded, companies I joined, a few that got acquired. The through-line is the same: how infrastructure really behaves once people depend on it.

Why systemctl daemon-reload drops GPUs from running Kubernetes pods

Thu, 17 Sep 2026 09:00:00 +1200

A systemctl daemon-reload on a Kubernetes GPU node can detach the GPU from containers that are already running. The pod stays Running, nothing restarts, and no component reports a fault. The workload simply loses the ability to open the device. Almost every package that ships a systemd unit runs a reload on install, so agent rollouts, apt-daily-upgrade and snapd refreshes all trigger it.

Resolving the Harbor uid 10000 vs FreeIPA admin collision on Kubernetes

Tue, 15 Sep 2026 09:00:00 +1200

getent passwd 10000 on a Kubernetes node that resolves FreeIPA through sssd returns admin. Harbor's Helm chart (1.19.2, app v2.15.2) hard-codes runAsUser and fsGroup to 10000 for every component except its Valkey cache. FreeIPA installed with --idstart=10000 hands that exact number to the first account it creates, the built-in admin, and gid 10000 to the admins group. Two sensible defaults, and a registry that would write every image layer to shared NFS storage as the directory administrator.

Handing one node's GPUs to HAMi without touching the live models

Thu, 10 Sep 2026 09:00:00 +1200

The box has four Blackwell RTX PRO 6000s in it. Two are Server Edition at 600 W, two are Max-Q Workstation Edition at 300 W. Kubernetes advertised four identical nvidia.com/gpu and handed out whichever card it felt like.

For a single notebook that is fine. For tensor parallelism across two cards it is not. A mixed pair paces every collective at the slower card, so a TP=2 deployment that lands on one SE and one Max-Q quietly runs at Max-Q speed forever. That was the blocker on a research AI platform we run for a New Zealand university: three GPU nodes, two of them carrying eight H200s each and serving a live model to researchers, one RTX node for development and benchmarking.

The Answer Was Already in Teams. We Just Couldn't Find It.

Mon, 07 Sep 2026 09:00:00 +0000

Somewhere in your company's chat history is the exact fix for the problem you are debugging right now. Someone wrote it down two years ago, in a thread, at 11pm. Good luck finding it.

That was the problem I set out to solve at ASI Solutions. Our knowledge was not missing. It was scattered. Decisions lived in Microsoft Teams. Delivery history lived in Linear. Deal context lived in HubSpot. Documents lived in SharePoint. Meeting notes lived in Granola. Proposals lived in PandaDoc. Invoices lived in Xero. Code and pull requests lived in GitHub.

How our document classifier votes before it ever asks a language model

Thu, 11 Jun 2026 09:00:00 +1200

A district council we work with has about 4 TB of files in SharePoint and more on internal SMB shares, and they need to know what is in them: which documents are contracts, which hold personal information, and which should have been disposed of years ago under the Public Records Act 2005. Every one of those files has to be read, classified into one of roughly 30 document types, and scored for sensitivity.

How I Built ASI's Marketing Engine Around AI Agents

Mon, 11 May 2026 09:00:00 +0000

This is a long one, and I mean that as a recommendation rather than a warning. Most of what gets written about AI in marketing right now skips the part that actually matters — the specifics: what broke, what surprised you, what you'd do differently. This piece doesn't skip it. If you're building something similar, or wondering whether any of this is real, I think you'll find it useful.

From XGBoost to LightGBM: How Our ML Model Adapted to Two Market Regimes

Sun, 19 Apr 2026 09:00:00 +0000

When we first wrote about SignalScope's ML backtesting pipeline in March, the model of record was XGBoost. Six weeks and hundreds of experiments later, it isn't. The current production model is a single LightGBM regressor with forty trees and depth two, trained on three-day forward returns, pulling almost all of its out-of-sample skill from a single feature. Mean information coefficient — the rank correlation between how the model ranks tickers and how they actually perform, where 0 is random and higher is better — has climbed from 0.006 on our first Ridge baseline to 0.161 today, a roughly 27x improvement. But the path between those two numbers wasn't monotonic. Twice in the last six weeks, a regime shift in the underlying data quietly invalidated our best model and forced us to start over. This is the story of how we rebuilt the pipeline three times — and what those rebuilds taught us about model stewardship in markets that keep changing underneath you.

Announcing the Ollama GPU Calculator

Mon, 28 Oct 2024 05:37:02 +0000

If you have spent any time running local models with Ollama, you already know the awkward part: picking a model is easy; knowing whether your GPU can actually host it is not. Parameter count, quantization, context length, and VRAM interact in ways that are hard to eyeball — especially once you start mixing cards or looking at Apple Silicon unified memory.

Roundtable: How to Bring Enterprise-Grade Security Capabilities to SMBs

Thu, 02 Dec 2021 12:28:58 +0000

Every day hundreds of organizations are breached in an attempt to steal corporate secrets and personally identifiable data. The most prolific attacks end up on the mainstream media. Still, countless data breaches go unnoticed by the media or even by the organizations themselves. I had the opportunity to be part of a roundtable on How to Bring Enterprise-Grade Security Capabilities to SMBs. SMBs also can have enterprise-grade security, Zero Trust, automation, and more without breaking the bank.

The end of zero-day VPN attacks with zero trust

Mon, 01 Feb 2021 15:31:19 +0000

Most organizations are still using VPN to provide access to internal applications and networks. Unfortunately, VPNs open the organization's doors to the entire kingdom (internal network). Even when ACLs, Rules, and VLANs segmentation is implemented, the chances of a bad actor executing a lateral movement attack are high, especially if there is a VPN config mistake.

How I automated my CI/CD pipeline with Google Cloud Run, Cloud Build and GitHub.

Tue, 19 Jan 2021 21:52:48 +0000

In case you are not aware, I develop and host a free widely used tool for sizing VDI deployments and also a few other tools. I have been developing for many years. Still, recently after migrating to Google Cloud, GoLang, and Docker containers, I have achieved what I consider to be a very automated and streamlined CI/CD pipeline that is extremely simple for anyone to implement for their projects.

Feature Spotlight: Citrix Secure Workspace Access – Part 3

Mon, 23 Nov 2020 23:06:47 +0000

In Part 1 of our series on Citrix Secure Workspace Access, we looked at why organizations need to embrace modern, consumer-modeled, user-friendly, and cloud-based working models, allowing choice and flexibility for BYO and modern SaaS applications. But at the same time, they need to ensure a safe and secure experience for external and hosted applications and data.

Feature Spotlight: Citrix Secure Workspace Access – Part 2

Thu, 12 Nov 2020 18:32:02 +0000

Cha-ching! My First Citrix blog post is now Live.

Tue, 03 Nov 2020 12:10:27 +0000

It has been one month since joining Citrix, and I am now getting into the rhythm.

When joining, I knew I would be responsible for Product Marketing for the Secure Workspace Access. Since then, I have also inherited the Identity portion of Citrix Workspace.

The Right Time for a New Challenge

Thu, 01 Oct 2020 11:06:19 +0000

End-User Computing has evolved since my time at the VMware EUC CTO Office, and it is now fast expanding into secure public-cloud and mobile-first solution stacks. Long gone are the days when VDI alone could cater for all remote and BYOD needs.

New VDI Calculator (HTML) debuts today!

Wed, 16 Sep 2020 05:30:36 +0000

Over the years, the VDI Calculator has been used thousands of times by architects and infrastructure teams to size VDI deployments. Just in 2020 alone, the tool has been used over 3,000 times, with the USA and Russia leading the charts.

VDI Calculator

Wed, 16 Sep 2020 04:58:47 +0000

Welcome to the (new) VDI calculator.

I am now retiring all old versions (Java and Docker) in favor of this new HTML version that can be used in desktops, tablets, and mobiles alike.

macOS “Must Have” Apps (Business & Geek) - 2020 Edition

Mon, 10 Aug 2020 11:44:18 +0000

In 2017 I released the list of my top apps for Mac users working in tech (not strictly). This is an updated version with a couple of additions and removals.

How VMware intends to Leverage Datrium DRaaS

Sun, 12 Jul 2020 14:05:14 +0000

VMware Announces Intent to Acquire Datrium to Provide Disaster Recovery-as-a-Service for Hybrid Cloud Environments https://lnkd.in/gREXsSD.

This video covers how VMware intends to leverage Datrium Disaster-Recovery-as-a-Service with VMware Cloud.

1 minute and 18 sec ... to DR a 13 Terabyte Microsoft SQL Server to VMware Cloud with Datrium

Thu, 18 Jun 2020 13:41:43 +0000

Yesterday I published Microsoft SQL Server Performance, Consistency at Peak, and The Art of Possible demonstrating Microsoft SQL Server with HammerDB TPC-C workload running on Datrium.

I thought that if I can run MSSQL so fast and consistently what would happen if my datacenter was suddenly gone and I used Datrium DR-as-a-Service with VMware Cloud. How long would it take for my 13 TB VM to come up online and be ready for my applications? (see my previous articles for additional information on VM and HammerDB)

Microsoft SQL Server Performance, Consistency at Peak, and The Art of Possible with Datrium

Wed, 17 Jun 2020 14:17:11 +0000

HammerDB is a free database load testing and benchmarking tool for databases. For this HammerDB test, I have chosen the TPC-C schema to build a roughly 500 GB database (to start with) of 200 warehouses and used 100 concurrent users to drive the database transactions with no think-time enabled.

Ransomware Gone in 1 minute with VMware Cloud

Sun, 10 May 2020 21:33:02 +0000

I hope you are all having a great day and a happy Mother’s day! Here is how I used the VMware Cloud on AWS to recover from a Ransomware infection in .... 1 minute ....! Some things must be seen to be believed!

Disaster! I moved my Citrix Farm to VMware Cloud in 8 minutes

Mon, 06 Apr 2020 16:13:36 +0000

I hope you are all having a great day at home. I recently posted on Twitter that I have executed DR from on-prem to the VMware Cloud on AWS of my entire Citrix Farm, including Active Directory, Delivery Controllers, VPN, and persistent desktops. As it got much attention, I decided to record a video; it's only about 4 minutes. Some things must be seen to be believed!

VDI is Now deemed Critical Infrastructure, but are Organizations Ready for it?

Sun, 29 Mar 2020 19:15:38 +0000

For a long time, EUC specialists and architects have been saying that VDI should be considered an integral part of the critical infrastructure and therefore given the same attention that core applications and databases receive in terms of availability and recoverability.

5 Reasons Datrium helps CIOs during an Economic Downturn

Tue, 17 Mar 2020 13:21:15 +0000

CIOs are advised to have plans in place in advance of a time of economic uncertainty or a possible downturn. However, IT organizations do not always have stable budgets to operate in a business-as-usual-mode, and in many cases, headcount reduction or deferral of IT projects is a reality.

Self-Inflicted Ransomware & Recovery in Minutes

Sun, 08 Mar 2020 19:43:07 +0000

The key to a successful recovery from ransomware is the combination of restoring a clean operating system from before the initial infection and then also recovering data from as recent as five minutes before payload activation.

I failed back my VMs from the VMware Cloud

Fri, 28 Feb 2020 00:37:11 +0000

Disaster Recovery is one of the most complex exercises IT will endeavor. From infra and data replication to applications and orchestration, it is hard! In this 2nd video, I show how to easily failback VMs from VMware Cloud on AWS.

Datrium Proactive Protection for Natural Hazards is now on Docker

Mon, 10 Feb 2020 17:20:30 +0000

I recently released a set of Python components that make Datrium take pro-active virtual machine snapshots and replicate them upon detection of high winds, high temperature, or earthquakes in a defined region.

I saved my datacenter in 4 minutes

Sun, 09 Feb 2020 18:52:52 +0000

DR is one of the most complex exercises IT will endeavour. In this video you will see how I executed DR for a couple of VMware VMs to the VMware Cloud in 4 minutes.

Here is why I copied the StorageReview Performance Benchmark for MySQL on Datrium... but there was a Gotcha!

Mon, 20 Jan 2020 09:47:38 +0000

If you want to benchmark a storage subsystem, it's best to use pre-existing benchmarks to be able to compare results. Instead of reinventing the wheel, I just copied the configuration and tests performed by our good friends at StorageReview. (https://www.storagereview.com/node/3598)

I solved Natural Hazard DR with Python and Datrium

Wed, 01 Jan 2020 18:31:30 +0000

I solved Natural Hazard DR with Python and Datrium.

One of the critical issues with disaster recovery, in general, is that you only want to trigger DR when the disaster strikes.

Backups - Feasibility vs. Requirements

Tue, 17 Dec 2019 20:56:51 +0000

A recent conversation regarding application protection triggered me to think about feasibility vs. business requirements. This topic applies to pretty much any enterprise technology domain, but in this case, I want to focus on backup and snapshot technologies. I'll use Datrium as an example given its high degree of flexibility.

We executed a Disaster Recovery Plan for our VDI!

Mon, 16 Dec 2019 12:40:04 +0000

I have been working with EUC solutions for a long time and for as long as I can remember, VDI solutions like Horizon and XenDesktop lacked a practical way to deliver DR.

DRaaS Connect gives LOVE to SAN/NAS and HCI with VMware Cloud Integration (...and Instant RTO)

Wed, 13 Nov 2019 10:37:16 +0000

Datrium announces integration with VMware Cloud for Disaster Recovery for customers using SAN, NAS, and HCI.

It's been fun writing about Datrium tech over the past few years as it unfolded as a cloud-first data management stack without leaving behind its on-prem components. Datrium is genuinely the multi-cloud solution that CIOs and Operators always dreamed of - a high-performance platform with integrated data management, backup, and Disaster Recovery policies that follow Apps wherever they go.

Bye Bye Java Web Start... Hello Docker Container and Direct Download says the VDI Calculator

Sun, 03 Nov 2019 20:02:23 +0000

Since I started tracking the VDI Calculator usage, back in 2013, there have been over 66,000 application launches, and since I released the current VDI Calculator back in 2013 I have been using Java Web Start to facilitate the launch of the application. Java Web Start allows you to download and run Java applications from the web with an easy, one-click activation of applications and it guarantees that you are always running the latest version of the application.

VMware Horizon now has DR to VMware Cloud on AWS

Mon, 28 Oct 2019 12:06:48 +0000

For as long as I can remember, VMware Horizon lacked a practical way to achieve DR, be it on-premise or in the cloud. To achieve DR, administrators had to abstract components of the solution, virtualize applications, manually replicate golden images and make heavy use of non-persistent desktops. Still, even like that, it was very complex to have an effective DR solution that cater for business requirements.

Innovators' Challenge - Normalizing the Clouds

Tue, 15 Oct 2019 16:35:45 +0000

The Old Lock-in Problem

Before VMware, there used to be three or four large block storage companies: IBM, EMC, HP, etc., and each produced their proprietary storage systems.

VDI Calculator now as a Docker Container

Thu, 10 Oct 2019 12:48:48 +0000

Over the years I've heard users of my VDI Calculator complain that they either have issues with Java functionality or they do not want to install Java on their computers due to security concerns.

Datrium DR-as-a-Service Networking with VMware Cloud (VMC)

Mon, 09 Sep 2019 10:42:32 +0000

During VMworld Datrium launched a unique DRaaS solution completely integrated into VMware Cloud on AWS. At a high-level DRaaS if a complete DR solution offered as a subscription that leverages the VMware Cloud and native Datrium cloud-based backups to deliver a fully-orchestrated low cost, low RPO and lower RTO disaster recovery for VMware customers.

World’s 1st Just-in-Time Cloud DR (VMware Cloud)... Everything Techies Need to Know....

Wed, 21 Aug 2019 09:26:23 +0000

No, the title of this post isn't a clickbait.

Organizations spend hundreds of thousands of dollars per year to keep Disaster Recovery (DR) datacenters operational while they wait for disaster to strike so they can realize that huge upfront and ongoing investment, but at the same time crossing fingers and hoping that the tragedy never happens. In many ways, DR datacenters are like insurance with huge upfront premiums.

VMware Cloud + Datrium = Awesome Recovery!!

Wed, 21 Aug 2019 09:17:09 +0000

VMware is the de-facto standard for the enterprise datacenter, and AWS has the broadest set of cloud services with global scale and reach. If you not aware, both companies have jointly engineered a cloud solution that delivers the best of VMware and AWS for customers. The VMware Cloud on AWS has a massive global presence.

What I have learned from 100 Oracle Benchmarks on Datrium

Sun, 07 Jul 2019 14:30:57 +0000

I have recently completed a series of benchmarks with Oracle on Datrium, and as it comes with no surprise Oracle is a perfect fit for Datrium's host flash acceleration and integrated data protection. I have documented tests, findings and recommendations in a document published here, but below you will find a few highlights and lessons learned.

ControlShift Series - Seamless Orchestrated VM Failback from DR Site

Fri, 14 Jun 2019 15:47:08 +0000

This is my 3rd post on this Datrium native DR series, and this time around, I demo the unique DR Failback of VMware VMs between sites and vCenters. Many solutions provide orchestrated failover, but Failback is the tough part and where most fail. See for yourself how Datrium handles it beautifully!

From Zero to Instant Orchestrated DR in 6 seconds. Live!

Thu, 13 Jun 2019 12:01:41 +0000

Hello! This is my 2nd post on Datrium native DR capabilities. This time around I discuss the initial DR plan, configuration, the failover, and the failback of VMware VMs between sites and vCenters.

How Datrium ControlShift and CloudDVX are Deployed and Operated on AWS - Part 1

Sun, 09 Jun 2019 23:24:15 +0000

As you may know the Automatrix platform include Primary Storage + Backup + Disaster Recovery + Encryption + Data Mobility in a single end-to-end solution.

Datrium Elevate NetApp Snap & Replicate to New Levels with Reverse Snaps

Mon, 20 May 2019 14:56:37 +0000

Multisite Replication Topology

In my article (Disaster Recovery Seeding is a Pain!) I write about Datrium universal fingerprinting coupled with over-the-wire deduplication to drastically reduce the amount of data transferred between sites when replicating. Data is always globally fingerprinted, deduplicated, compressed and erasure coded, and data blocks are uniquely identified and logically aggregated. The system holds a full understanding of existing and missing blocks of all sites. However, depending on the data change rate, link and site availability there could be replication issues. Let's explore.

Datrium 5.0 Features Overview and Big Welcome to Automatrix (Beyond Marketing)

Wed, 15 May 2019 10:18:15 +0000

It’s that time again… Datrium is dropping a new release with awesome features! This is one those long posts… be patient, get a coffee and read through, or simply go to the section that you are interested in.

Datrium is gaining market share, and FUD escalates to new levels. Here I technically dismiss legacy vendor FUD against Datrium.

Thu, 09 May 2019 13:38:46 +0000

No matter what, when up and coming technology vendors start to disrupt legacy players, the amount of FUD (fear, uncertainty, and doubt) broadcasted by such vendors increase. Datrium being the major disruptor to the legacy storage and HCI business make competitors starting to spread ‘misguided’ information about our technology.