The problem with cloud services is that you put all your eggs in one basket. Even if outages are less frequent, impacting more people at a time isn’t good. If most people use a handful of centralized services, those services become a larger target for hacking and DOS attacks.
That’s why I like on-prem, generally speaking. It localizes the risk and prevents a cascading effect.
Theoretically the major cloud providers like MS have redundant geographically dispersed servers that mean there should only be an outage if the individual user can’t reach the internet.
In practise however those promises are hollow for a number of reasons, cost usually. Legal issues like GDPR also impinge (EU data being allowed to be in the US has been blocked by the courts the other day for example). In addition there’s a long list of other configuration reasons which almost always come back to cost indirectly.
Theoretically an ideally configured cloud solution is far superior to on-prem.
In the real world, not so much: corners cut, pennies saved by non technical managers not understanding the ramifications of their choices & etc
On prem is certainly better in the real world if you’re big enough to afford proper redundancy and to hire and keep good techs.
Many many firms can’t tick those boxes though and so you get to imperfect world optimisation where what is good for coy. A is bad for coy. B
I’ll add one more: privacy. The more people rely on a given service, the more juicy it is to attack. On prem limits the attractiveness of your data, so you’re hiding in a crowd instead of trying to protect a single golden goose.
The problem with cloud services is that you put all your eggs in one basket. Even if outages are less frequent, impacting more people at a time isn’t good. If most people use a handful of centralized services, those services become a larger target for hacking and DOS attacks.
That’s why I like on-prem, generally speaking. It localizes the risk and prevents a cascading effect.
Theoretically the major cloud providers like MS have redundant geographically dispersed servers that mean there should only be an outage if the individual user can’t reach the internet.
In practise however those promises are hollow for a number of reasons, cost usually. Legal issues like GDPR also impinge (EU data being allowed to be in the US has been blocked by the courts the other day for example). In addition there’s a long list of other configuration reasons which almost always come back to cost indirectly.
Theoretically an ideally configured cloud solution is far superior to on-prem.
In the real world, not so much: corners cut, pennies saved by non technical managers not understanding the ramifications of their choices & etc
On prem is certainly better in the real world if you’re big enough to afford proper redundancy and to hire and keep good techs.
Many many firms can’t tick those boxes though and so you get to imperfect world optimisation where what is good for coy. A is bad for coy. B
Sums up my thoughts pretty succintly.
I’ll add one more: privacy. The more people rely on a given service, the more juicy it is to attack. On prem limits the attractiveness of your data, so you’re hiding in a crowd instead of trying to protect a single golden goose.