One chapter ends and a new one begins

From our own hosting platform to the Microsoft Azure Cloud

Why we are turning everything upside down again after ten years of high availability and what our customers get out of it. In 2014, our Citrix platform was the Ferrari of SME data centers: 1,000 mailboxes, 700 desktops, bank-level security. In 2025, the rules of the game will be different: Zero trust models, AI defense and global geo-redundancy will emerge first in hyperscalers. This is precisely why we are moving our entire infrastructure to Microsoft Azure. In this article, we show how we prepared for this step, why an in-house data center will barely be able to keep up in 2025 and what tangible added value our customers are now experiencing on a daily basis: more security, higher availability, limitless scaling and clearly calculable costs.

1 | Prologue - ten years of innovativedrive

When we equipped the first racks in our Zurich data center in 2012, we had an ambitious plan: SMEs should experience the same high availability, performance and security as enterprise groups, only without an enterprise budget. Two years, countless whiteboards and a few night shifts later, the time had come: in spring 2014, our own multi-tenant Citrix hosting platform went live.

Today, over a decade, thousands of operating hours and several generations of hardware later, we are taking the next logical step and migrating everything to Microsoft Azure. In this article, we take you with us on this transformation and explain why "in-house sheet metal romanticism" is technically exciting, but strategically outdated, especially from a security perspective.

2 | Review - The hour of birth

For technology enthusiasts who lovedetails

Between 2012 and 2014, we implemented the Citrix Reference Architecture for Multi-Tenant Hosters in its purest form and consistently separated the basic infrastructure (Hyper-V / System Center VMM) from the cloud layer. Our hybrid model combined shared services such as Exchange, SQL and file services with fully customer-dedicated resources. All core services were running active/active: a two-node Exchange DAG, SQL and SMB guest clusters, dual Citrix NetScaler load balancers inside and outside, and a Barracuda F800 firewall cluster as the central switching point.

The compute layer consisted of twelve Nutanix HX nodes, each with 384 GB RAM and 28 CPU cores (56 threads), resulting in a total of over 300 vCPUs, 4.5 TB RAM and 140 TB usable RF2 storage. Virtualization was carried out on Windows Server 2012 R2/2016/2022; the applications were provided via Citrix XenApp 7 LTSR for around 700 simultaneous sessions, the endpoints ran on IGEL thin clients. We automated patch and third-party update cycles via SCCM with depot integration, while SCOM provided 24×7 telemetry and escalated immediately via on-call support. This was supplemented by PowerShell DSC scripts and Automate101 workflows for policy and GPO handling.

The platform was hosted in the Interxion data center in Zurich with full ISO 27001 and FINMA compliance, its own PKI environment, and all backups were stored in an off-site target. More than 1000 Exchange mailboxes, 700 VDI / XenApp sessions and an SLA of 99.95% underlined the enterprise claim - delivered at an SME price. For 2014, this was a true "Enterprise-as-a-Service" environment; however, increasing zero-day frequencies, NIS2 / DORA regulations and the innovation rate of hyperscalers make having your own data center seem less efficient today than the logical step towards Microsoft Azure, which we are now taking.

For all those who didn't understand anything, here's some more without "IT slang"

Imagine if we had started building a futuristic apartment building for data in 2012. We worked on it for two years until we finally opened the doors in 2014. Each company that moved in was given its own lockable apartment, its own personal piece of digital privacy. The common rooms were comparable to the laundry room, heating and emergency power generator and were our central email, backup and security systems, which were shared by everyone. We installed two elevators so that nobody got stuck in the stairwell: If one failed, the other kept running so that operations never came to a standstill. Twelve powerful "engines" were humming under the foundations - ultra-modern Nutanix servers with more computing power than a medium-sized company would ever need on its own. This allowed over 700 employees to work simultaneously on their virtual Windows desktops without a hitch, and more than 1000 full e-mail inboxes remained as relaxed as a refrigerator in winter.

All of this was not in some basement, but in a high-security building at bank vault level: cameras, locks, access controls, ISO 27001 and FINMA certificates included. For the worst-case scenario, we also stored all data in a second, physically separate data center. Our customers didn't have to cool server rooms or plan night shifts, they simply moved in with us and enjoyed enterprise-level technology at a mid-market price. It was a real showcase project at the time, but technology continues to run marathons, and if you want to stay at the top, you have to keep up.

3 | The turning point - why local data centers are reachingtheir limits today

At some point, we realized that although our gleaming machine park looked impressive, it was increasingly becoming a drag. Every five to seven years, the natural hardware lifecycle forced us to undertake costly "lift and shift" projects: new servers, new storage shelves, new network cards and the same stressful relocation every time, just so that the foundation was up to date again. As soon as the fresh hardware was in the rack, the zero-day gaps were already chasing us through the weekend. In the cloud, patches are almost invisible; on-prem, they mean night shifts because we can't open maintenance windows in the middle of the day.

At the same time, regulation exploded: NIS2, DORA, the revised Swiss Data Protection Act - all important, all right, but in our own basement, every new directive costs additional compliance budget. While we were still planning ISO audits, cyber criminals had long since switched to global scaling, automated attacks with AI and jumped effortlessly across national borders. It's hard for a local business to keep up. And finally, the pace of innovation: The first productive AI services, confidential computing instances and even quantum-safe cryptography are now being created in the hyperscale cloud, far before they end up as a box on the server shelf.

The better we polished our old platform, the clearer it became to us: our mission is to empower customers, not to herd shiny tin. That's why we let go of the screwdriver and reach for the scaling controller in the cloud.

4 | Why Microsoft Azure? A securityreality check

Before we finally set off in the direction of Azure, it's worth doing a sober reality check: How does your own, lovingly maintained data center fare in a direct duel with a global hyperscaler? The comparison below shows where the differences lie, from physical access control and patch management to budget and compliance.

Spoiler: What used to be a competition at eye level is now David versus Goliath, except that thanks to the cloud, David no longer has to do any fighting himself.

In short: a single medium-sized company can no longer realistically provide the same level of security as a hyperscaler and, thanks to the cloud, doesn't have to.

5 | Security first - the five Azure arguments thatconvinced us

In the IT world, it has long been the case that protecting data protects business. But while traditional data centers function like solid lockers, Microsoft Azure is more reminiscent of a high-security vault with facial recognition, its own bodyguards and self-healing walls. This is exactly what convinced us. Five reasons stand out in particular, each strong on its own, unbeatable together.

Zero trust by design

In Azure , distrust is the basic principle - towards every device, every user, every application.

• Identity as the new perimeter: If you want in, you have to prove your identity; the "lock" is no longer on the office door, but in front of every service.

• Conditional access: access is approved according to the situation and depending on location, device, time of day or risk score.

• Just-in-Time & Privileged Identity Management (PIM): Admin rights are only granted for minutes and exactly when they are needed and disappear again automatically. Result: Attackers hardly ever find permanent keys because there simply aren't any.

Intelligent defense

Azure sees into the distance before the storm arrives.

• 65 trillion telemetry signals flow into the Microsoft Security Graphevery day. This identifies patterns that would be completely invisible to individual companies.

• 600 million fended-off attacks per day feed the AI models - every new trick used by attackers ends up in all protection mechanisms within minutes.

• Automated playbooks block compromised accounts, stop ransomware rollbacks or isolate suspicious VMs, often before the first human reads an alert.

Verified compliance

Instead of passing every audit from scratch ourselves, weinherit Azure's seal of approval.

• Over 100 certifications are available "out of the box", including ISO 27001, FINMA, GDPR, HIPAA or NIST.

• Audit trails, protocols and log integrity are predefined; auditors receive ready-made reports and we save weeks of verification work.

• New regulations often end up in the Azure controls first; what will be mandatory in Zurich tomorrow is often already implemented in Redmond today.

Resilience & geo-redundancy

If you hate downtime, you build in breadth and Azure builds in continents.

• Three availability zones per region mean that power, network and cooling are physically separated. If one data center fails, the second takes over, then the third.

• Cross-region disaster recovery automatically replicates workloads to a completely different region with just a few clicks.

• Thanks to live migration, even kernel patches remain invisible to users; VMs are moved before the screw is tightened.

Future-Proof Services

Security is a race: Azure provides us with the starting blocks of the future.

• Confidential VMs encrypt data not only on disk, but also during computation in memory.

• Post-quantum cryptography is being prepared long before quantum computers are available in production labs.

• Security copilots based on generative AI summarize log floods in plain text, suggest defensive steps and automate routine checks.

6 | Our cloud journey - this is how the switchhappens

Our journey to the Azure cloud was not a wild ride, but a precisely planned adventure: first, we checked every single workload like a travel agency checks the passports of its guests, maturity level, costs, risks. Meanwhile, a fully automated and standardized environment grew in Azure, complete with security policies and monitored network traffic. Then the actual move began: we lifted older virtual desktops into Azure Virtual Desktop almost unchanged, took databases and files with us, modernized them en route and let them fly on to PaaS services. As soon as we landed, tags and budget alerts ensured that no one got into the wrong cost trap, while our 24/7 desk guarded the new home and applied patches without night shifts. Everything ran in parallel, failover tests included, and the users didn't notice anything about the change of scenery, except that everything suddenly went a little faster. Now that all the customer apartments are set up in the Azure penthouse, we are just tightening the final screws and enjoying the view from the cloud.

7 | What does this mean for our customers?

What does this mean for our customers? Quite simply, their data is now stored in the best-guarded vault in the world without them having to invest a single franc in new hardware. Core services run with 99.99% availability guaranteed by Azure across multiple zones, and outages practically disappear from the scene. If a customer suddenly needs twice as much computing power because a major project comes knocking, one click is all it takes for capacity to grow in minutes rather than months. At the same time, state-of-the-art tools such as AI assistants, real-time analytics and automated workflows are available "on demand" and are just waiting to accelerate business processes. And all this at calculable pay-as-you-use costs: you only pay for what you actually use instead of swallowing a lump of investment every five years. In short: more security, more speed, more innovative power, without a budget hangover.

• More security: You benefit from the world's highest security standards without having to make your own investment.

• Higher availability: SLA-backed 99.99% for core services, multi-zone redundancy.

• Scaling in minutes: Growth or project peaks are covered with a click.

• Innovation on demand: AI features, analytics & automation are available immediately.

• Calculable costs: pay-as-you-use instead of block investments every five years.

8 | Conclusion

Our Citrix platform was a milestone that shaped us as Dinotronic. But true excellence means letting go of the good to create something better. With Microsoft Azure, we are consistently focusing our portfolio on maximum security, availability and speed of innovation, and thus on the benefit of our customers.

Would you like to know what your own journey to the hyperscale cloud could look like?

👉 Contact us for a no-obligation cloud journey assessment or visit our Managed Datacenter Service page.