Amazon cloud outage knocks out Netflix Pinterest and Instagram, or does it?

While the report here;

http://www.cloudpro.co.uk/cloud-essentials/3993/amazon-outage-knocks-out-netflix-pinterest-and-instagram?utm_campaign=itpro_newsletter&utm_medium=email&utm_source=newsletter

is undoubtedly true and factually correct, in that recent storms caused issues with Amazon’s data centre in Ohio, and previously they have had issues when their data centre in Ireland was damaged by lightening, the question should be what could be done differently, rather than ‘cloud services are not robust / safe.

I am a firm advocate for insuring you understand your contract with your cloud provider with and that you pay great attention to things like SLAs and guaranteed uptime.  This is especially true if you are using SaaS or PaaS type services that may in turn rely on another vendors IaaS service – you need to understand the layers to ensure your provider is not offering SLAs that it cannot meet due to them being more stringent than those of the providers of the services on which it relies.

However I question why this is considered an issue particular to ‘cloud’ based services.  These same issues could happen to any co-location / data centre hosting solution, and these along with many more minor issues are likely to cause disruption to anything you host locally in your server room no matter how grand a name you give it.  Sorry that’s one of my other pet hates, businesses with small server rooms that insist on calling them ‘data centres’ or other grandiose names and talking about them as if they are a large and resilient as actual Data Centres etc.

Anyway, back on topic, obviously when a cloud service provider has an issue it is likely to affect many customers so will be news worth, but before you worry too much or begin to dismiss the idea of moving some or all of your service to the cloud, ask yourself is it likely to be more or less robust than hosting things yourself?

Take the necessary precautions;

-Understand the offering you are purchasing, the SLAs and guaranteed uptime in the contract,

-Build BC and DR into your service; ensure it is replicated to multiple servers and disks locally, and to another geographically disparate data centres and you can host a hugely robust solution in the cloud.

K

3 thoughts on “Amazon cloud outage knocks out Netflix Pinterest and Instagram, or does it?”

  1. Nice write up Kevin. You’re spot on – cloud doesn’t fix all the problems. You still need to consider the full service and make sure BC and DR are part of it. And look at all the what-if scenarios. Even more critical for SMB’s who might not have the same level of internal technical resources.

    1. Totally agree, I was also trying to highlight that ‘cloud’ shouldn’t be excessively blamed for outages as these occur in pretty much all scenarios and likely more often for companies that try to run their own IT with considerably less resilience and IT expertise. This is why contracts and things like SLAs are key when moving data and / or systems to the cloud.

Leave a Reply to snowballsinwinter Cancel reply

Your email address will not be published. Required fields are marked *