Managed cloud: the right question is not where the servers are

Postazione di lavoro remota su laptop: desktop e applicazioni pubblicati dal cloud gestito

In many companies the server cabinet sits in a room that was never meant for it: a storeroom with no dedicated cooling, a UPS bought six years ago, a few cables labelled in pen. It works, and that is why nobody checks it. Those who start assessing managed cloud rarely do so after a serious failure. They get there because they have to replace an end-of-life server and find they have no answer to a simple question: how long would the company be stopped if that machine went down on a Tuesday morning?

The conversation almost immediately moves to the physical location of the data, which is the least interesting part of the problem.

What matters is who answers when the servers stop

Two managed cloud contracts with the same name can describe very different services. In the first, the provider supplies compute capacity and space, while operating systems, updates, backups and access management remain with the company. In the second, the provider covers the whole chain and answers for it working. The difference shows the first time something breaks, when nobody knows who should step in.

That is why the first page to read in an offer is the responsibility matrix rather than the list of resources. If that table is missing, ask for it before discussing the fee.

Operationally, the comparison between on-premise servers and managed cloud comes down to a few items:

Operational item On-premise servers Managed cloud
Updates and patches Handled by internal staff in spare time Scheduled by the provider in agreed windows
Hardware failure Spare parts and repair to be sourced on the spot Redundancy built into the infrastructure
Restore tests Postponed when time runs short Scheduled, with a written report
Night and holiday cover Dependent on individuals’ availability Organised in on-call shifts
Capacity growth New purchase at every jump in demand Resources extended on the existing environment

Two sites, everything in one cabinet and one person who knows how it works

In an analysis at a company with two connected sites the picture was this: an internal mail server, an internal terminal server, ERP and document archive on the same perimeter, and a single entry point from outside through a VPN to that terminal server. Traffic stayed inside the company network, apart from web browsing.

None of those elements was wrong in itself. The risk lay in the sum: a failure of the network device at the main site also stopped the second site, and only one person knew the configurations, an external technician who came when called. During his holidays the company had no workable alternative.

The cloud platform drove the solution. With management we drew up the list of services that cannot stay down for half a day, and the migration to our managed environment followed that order: first email and terminal server, replicated and reachable from both sites even with the main site down, then the rest according to a written schedule. Configurations that lived in one technician’s memory are now documented on the infrastructure, and support no longer depends on one person’s holidays.

The signs that make a move like this timely can be recognised in pictures like this one: hardware that needs replacing anyway, absences that are hard to cover with internal resources, several sites to connect, a recovery plan still missing, and customers asking for contractual evidence on data handling.

In managed cloud, continuity is assessed beforehand, not during the failure

Many discussions confuse backup with business continuity. A backup answers the question of recovering data. Continuity answers a different question: how quickly the company gets back to work. The two have different architectures and costs, and a company can legitimately decide to tolerate half a day of downtime on some services and not a minute on others, as long as the choice is conscious and written down.

Our managed cloud platform runs on our own Tier IV data centres in the European Union, with the infrastructure replicated to a second, geographically separate site. We are a Nutanix Professional Partner and are building the new generation of the platform on that technology. The organisation that delivers the service is certified ISO 9001, 27001, 27017 and 27018.

Certifications say something, not everything

A certificate does not solve a project and does not replace acceptance testing. It does show a fact that a third party can verify: administrative access management, environment segregation and personal data processing follow controlled procedures rather than internal habits.

In managed cloud, remote access is the change people notice

Publishing remote applications and desktops affects users’ day more than any infrastructure choice. People open the ERP, the CAD or the accounting program and work, while the data stays in the data centre and not on the laptop travelling by train. The model covers home workers, branch offices and external consultants, who reach only what falls within their remit.

The item to design with the same care as the servers is each site’s connectivity: well sized, it makes remote work indistinguishable from office work. An undersized link turns a sound architecture into a daily source of tickets. It is worth measuring actual bandwidth, not the nominal figure in the contract: on more than one site visit we found a fibre line already active and never connected to the equipment.

Questions to ask a managed cloud provider

Ask a managed cloud provider in writing about data location, operational responsibilities, tested recovery times, what the fee includes and exit terms. Before signing, not after.

  • Where the data physically resides and where the backup copy is.
  • Who administers the operating systems and who applies patches.
  • What recovery time the provider commits to, and for which services.
  • How often a real restore test is carried out and who receives the report.
  • What the fee includes and which activities are billed on consumption.
  • How to exit the contract: in what format the company gets its data back and how quickly.
  • Which certifications cover the service being sold, not just the company selling it.
  • Who answers at night and on public holidays, on which channel and with what response time.

The second to last point escapes almost everyone. A certificate in the provider’s name does not by itself cover the data centre of a third party delivering the service.

With managed cloud, whoever leads IT keeps their role, if the scope is written down

A recurring fear of an internal IT manager concerns their own role: if the infrastructure moves to a provider, what is left to do? In the projects we run the opposite happens. Routine maintenance and on-call shifts pass to us, while the internal person goes back to processes, integrations and relations with departments, the part of the job no provider can do in their place.

For the model to work the scope must be defined in writing: who decides a configuration change, who authorises new access, who opens a ticket and with what priority. At Aesir Srl we keep fixed-fee IT management and managed security within the same contract, as MSP and MSSP, because when an incident involves both infrastructure and security the customer should not have to coordinate two companies. On business continuity the method is described in the guide to business continuity management with Nutanix. We work on projects across Italy and, on specific request, in Switzerland, the United Kingdom and the United States: every migration has left a repeatable procedure and a list of mistakes not to repeat.

Frequently asked questions

Do the programs in use need replacing?

In most cases no. ERP systems, document applications and vertical software run in a published environment without rewriting. The exceptions are very old applications tied to local peripherals, and they are checked during the analysis.

What happens if the internet connection drops?

Workstations connected to centralised services stop, just as they would for a failure of the internal network. For sites that cannot tolerate interruptions, a second connection on a different carrier is provided, with automatic failover.

Does the data stay in the European Union?

Yes: both the primary environment and the replicated copy reside in the European Union. It remains a point to check in writing with any provider, because it affects the company’s obligations towards its own customers.

How long does a migration take?

A migration to managed cloud depends on the number of applications and the tolerance for downtime, not on the gigabytes to move. The long phase is parallel testing, when the old environment stays on while users try the new one on real work.

Let’s talk

Anyone assessing managed cloud can start from an internal exercise without involving providers: list the services the company uses every day and write next to each how much downtime is tolerable and who would physically restore it today. The rows left without a name next to them describe the real extent of the risk.

If you would like to explore the topic or assess your company’s situation, fill in the form at the bottom of the page or write to support@aesir-tech.it: we will book a free consultation and start from your numbers.

Vuoi parlarne applicato alla tua azienda?

Ti rispondiamo con l’esperienza di progetto, non con una brochure.

Request a call