It's very true, no cloud provider is available 100% of the time. It's predominately the reason why in the industry we try to utilize multiple cloud providers to ensure less chance of a systems outage. This also protects against internet outages as well, which happen from time to time.

Looking at cloud providers as a whole, while outages do occur from time to time....rarely, if ever, is any data actually lost. If you compare that with companies who do things internally, how often do many of them have 100% uptime or 0 loss of data.. For example, when has a company ever found that a system failed and their backups didn't work and they lost data....this happens with some regularity. Or if a system fails midday, they have to revert to a backup from a previous point in time and they lost X number of hours of data.

Hi there

@pparks1

Multiple providers definitely help but one has to then consider data security -- if backups can be restored from various sites and providers that also opens up a whole new can of worms as regards security etc.

It just seems as some of those journos suggest that the more we depend on these sorts of services - outages from whatever reason and however small are going to cause massive disruption in how business etc is conducted today and we need to be mindful of the consequencies if we don't have other emergency systems in place to continue commerce while the main service is being repaired / resumed.

Perhaps these sorts of lessons are best learnt these days before we are 100% reliant on these sorts of providers so we can have systems in place in the case of these things going down either via wilful hacking or just Software glitches (or both).

We've probably been too complacent on "The Resiliency" of the Cloud" - which so far hasn't been that bad actually - but the more you rely on it without some alternative strategy in place if part of it fails then the worse the effects will be when the inevitable happens.

Data Loss isn't the only problem here -- if a business for example relies on real time transactions -- e.g FOREX (Foreign Exchange) or other types of transactions --particularly in Stocks etc - Data Loss isn't the main issue -- restoring these systems back say a few hours won't lose data but the amount of trading losses could run into trillions of dollars since the actual online real time trading would not have been captured since the last backup.

Cheers
jimbo

jimbo45 said:

Multiple providers definitely help but one has to then consider data security -- if backups can be restored from various sites and providers that also opens up a whole new can of worms as regards security etc.

Absolutely true, hence the reason that as an enterprise that relies heavily on AWS, my organization encrypts everything. EBS volumes on EC2 instances are encrypted at rest, databases themselves are encrypted at rest, everything is https and encrypted in transit, etc.

jimbo45 said:

It just seems as some of those journos suggest that the more we depend on these sorts of services - outages from whatever reason and however small are going to cause massive disruption in how business etc is conducted today and we need to be mindful of the consequencies if we don't have other emergency systems in place to continue commerce while the main service is being repaired / resumed.

Yeah, the cloud is a totally different mindset.

In the old days, we racked and stacked servers and we troubleshot problems on the servers and we backed up the data on the server so that we could restore it. We maintained systems for years and had extensive documentation. When a server went down, the techs were paged and we logged in to fix the issue.

Now, we build everything using infrastructure as code. Server are deployed via this code, and the data on the server itself is not critical. When a server goes down, the instance is simply terminated, and a brand new image is spun up to take the place of the broken server. We don't often troubleshoot a problem on a node anymore, we just check logging in an enterprise tool like Splunk. We can easily roll out new code multiple times per day, and if a problem exists, we simply roll the change back out. And 90% of the time, we don't use VM's anymore either, it's all containers and Kubernetes.

jimbo45 said:

We've probably been too complacent on "The Resiliency" of the Cloud" - which so far hasn't been that bad actually - but the more you rely on it without some alternative strategy in place if part of it fails then the worse the effects will be when the inevitable happens.

Yes, that's why the cloud providers replicate your data to multiple data centers, and have geographical presence in multiple countries all over the globe. And also the reason why organizations like to have multi-cloud presence too to protect against that random outage.