What’s your plan when the cloud fails?


Over the last few years, cloud computing has moved into the mainstream.  More and more companies are adopting ‘cloud-first’ initiatives and moving more infrastructure over to public clouds, such as AWS and Azure.  These clouds should not be underestimated – they are such powerful tools that even entirely on-prem solutions are using the agility and flexibility they offer.  This agility – the ability to say ‘I want a VM’ and have one running in minutes – helps keep IT users in house and avoids shadow IT springing up.  The flexibility allows applications of all sizes and use cases to run from a single pane.  Companies are able to move from "thought" to "action" faster then ever.

With all of the cloud goodness available, it’s no wonder that AWS grew 63.8% last year.  Even with this, a public cloud is not a panacea.  It’s not a replacement for your IT paradigms, and it’s certainly not a replacement for proper planning.

There is no Cloud

All of your storage and compute resources have to come from somewhere.  Even in the cloud, it’s just hardrives and x86 servers.  This means that, just like infrastructure today, failures are not an if but a when.  Hard drives fail.  Databases get corrupted.  Server hardware fails.  The ‘cloud’ simply takes all the compute and storage resources, bundles them up, and then carves them out with a far easier to use GUI than traditional tools.

Amazon EC2 volumes, for example, have a 0.5% chance of failure.  Amazon themselves recommend backing up your data to another datacenter to protect against this failure.  SFDC lost 4 hours of customer data earlier this year.  Even Office 365 only has 99.98% uptime over the last few quarters, which at first glance seems pretty good.  However, doing the math this translates to almost 30 minutes of downtime over a single quarter, or 2 hours over a year.  Not the end of the world, until you're in the middle of it.

That’s not to mention maintenance.  With your on-prem environment, you can plan for downtime – no one knows your business better then you.  With a cloud environment, you’re beholden to them and their schedules. 

If you were one of those unlucky customers with data loss with Salesforce, did you have a backup?  Just because there’s a cloud environment doesn’t mean that the 3-2-1 rule for data that you care about stops applying (3 copies, 2 vendors, 1 remote site).  Since you can’t see the infrastructure or failure domains you can’t count a public cloud as anything more than a single copy.

If you want something more customized, there are plenty of Service Providers out there who can cater a solution to your particular needs – making sure your cloud of choice matches your business, rather than forcing your business to match a particular cloud.  Using one of these providers, you can customize your solution with higher uptimes, better guarantees, or just simple a more favorable maintenance schedule in a way that would be impossible with the massive clouds.  These hosted (or virtual) private clouds are simply another way to consume cloud resources.

Using the right tools you can even give your users this experience with on-prem private clouds.  Enterprise Cloud solutions are a great way to do this - simple, linear scaling gives your infrastructure a public-cloud like agility, but allows you to maintain the control and security that you need to run your business with confidence.

None of this changes that public clouds, hosted private clouds, and Enterprise Clouds are powerful IT tools that free your high value employees to focus on other things, rather than the minutia of the infrastructure.  It’s important to keep in mind, however, that any IT tool has tradeoffs and you need to note these tradeoffs to shore up their weaknesses with other solutions so you’re not left with a nasty surprise one day.

So what’s your plan when the cloud fails?