Everything should be on a cloud

Posted by Wayne Sheddan on 11 April 2016

As something of an infrastructure architect, I have come to realise that all workloads should be on a cloud. Now, if what you just heard me say seems to say that everything should be on a public cloud, and your ERP/CRM should use the vendor SaaS platform I believe you have seriously misunderstood cloud.

I have come to realise that cloud has NOTHING to do with the location of IT. Rather it is HOW IT is done. If for you, going cloud is moving your current mess to a remote set of data-centers, or to use the software you have no control over, all you have really done is spread your mess all over the world.
But if the cloud is changing how you fundamentally use IT to a more standardised set of components, with automation and using applications abstracted through an interface virtualization layer (API gateway, Webservices, ESB) you are on the right path.

Too often I see the 'going cloud' thing is NOT accompanied by a fundamental change in application taxonomy. Instead going cloud is often just a lift and shift to a location of unknown provenance, to a distant location with system integrations becoming a global spaghetti of API calls and ETL. What was once a nice consolidated mess becomes a security, networking and compliance nightmare. Often this smearing of processing to all corners of the world is done with little regard for things like integrations, availability, disaster recovery and flexibility. And it seems that often going public cloud is done in the hope its cheaper and easier than what is currently in use.

Well, guess what?
Like most worthwhile things in life IT is neither cheap nor easy – no matter how you do it.
And like almost everything in life if it is cheap and easy its usually not very desirable, or will make you sick. And when it comes to being cheap – how many have an actual handle on what the on-premise solution costs? And without that how do you actually KNOW going public cloud is cheaper?

Returning to my premise: go all in on cloud – the benefits of standardisation and introduction of stateless middleware providing services for systems of engagement to utilise are worthwhile.
But make sure you think cloud as being How you do IT – NOT Where. Public/Hybrid or on-premise – cloud is a great way to improve IT service.
But you need to do a whole IT transformation as you take on the increased levels of discipline cloud requires. It is this change in a discipline that really drives the benefits of cloud – benefits not dependant on where the cloud is. Public/hybrid or on-premise: it is the rigour of cloud that drives the benefits.

Additionally, make sure you do entire system apples for apples cost comparisons:
For example, if the public cloud provider does oversubscription of CPU, memory and disk storage (they all do – aggressively!) then compare that with an equivalently oversubscribed on-premise environment. Remember to add in network costs, security costs, backup costs, AV costs, monitoring costs,… that $30/month public cloud VM cost pretty quickly escalates. And it gets worse when you do something sensible like a backup to another cloud environment and get slammed with the data egress charges. Remember a backup isn't a backup if it is on the same storage platform as primary, and most cloud providers use the same underlying storage platform for everything – including what they call 'backups'.

I also ponder on how moving to the public cloud seems driven by a nearly universal desire to avoid CAPEX. I wonder whether this reflects a lack of confidence in NZ's economic future… because a 3+ year TCO sure doesn’t point towards public cloud very often. Or is this just symptomatic of our cultural obsession with reducing costs – without regard for the future. This is a concerning aspect of the current public cloud hypergasm – what it says about our cultural willingness to commit, or NOT as the case seems to be. 

Just as frightening are the long term repercussions with our cost reduction obsession, rather than the long-term focus of TCO and ROI. Focus on cost reduction is to join the race to the bottom – focus on TCO and ROI is to focus on climbing to the top. I know where I'd prefer to be.

Wayne Sheddan is Storage Practice Lead at Open Systems Specialists. Wayne has over 25 years in the IT Industry and specialises in Storage consulting and architecture.