Thursday Read: The Cost of Cloud

exa-Chris
Community Manager
Community Manager

The move to the Cloud has been THE most driving force behind many current IT Projects. Understandably as the promise is faster, easier, cheaper. "Cloud or not to Cloud or what else?" was the topic of a presentation that we gave two years ago during the TDWI in Munich. So as time ticks and with CoVID speeding many things up, we should all be settled and fine?

A few days ago I came across this article in VentureBeat. It discusses the move to the cloud and the partial way back for some analytical cases. This reminded me of so many changes in the IT, centralized then departementalized and then back again full circle.

You can read the article here: 

https://venturebeat.com/2021/06/04/the-cost-of-cloud-a-trillion-dollar-paradox/

I was wondering, what’s your take on this topic and where are you heading? Cloud All In or not, or hybrid of some sorts?

Regards
Christian

Connecting Customers, Partners, Prospects and Exasolians is my passion. Apart from that I cycle, listen to music, and try to understand what all those technical discussions really mean...
2 REPLIES 2

jwarlander
Xpert

It's an interesting article!

To me, it certainly means you should keep your eyes on the numbers, and probably consider hybrid workloads the default once your infrastructure starts stabilizing even a little bit.

The article discusses mostly how to approach the issue from a company-wide perspective, but I'd like to think that it's probably more relevant to bring to a team-and-project level. Even if the company as a whole is stable, perhaps an internal backend system is being redesigned, and will go through a lot of frequent changes until anything is certain about the infrastructure?

The big question is probably how to work with infrastructure in a way that easily transfers between cloud and on-premises. Infrastructure-as-Code approaches most likely won't be identical, but can still work in similar ways if you utilize things like OpenStack, Kubernetes etc in your on-premises deployment.

Naturally, it also means thinking about the pros and cons if you start adopting many provider-specific solutions for serverless deployments etc; which of those can easily transfer to an on-premises option, and which ones would be harder to migrate?

Great food for thought!

exa-Graham
Team Exasol
Team Exasol

All good points - and indeed 'food for thought' and many sentiments echoed by Matthias Golembek here: Big data at 10: did bigger mean better? - Exasol 
There are many currents, opposing forces and hype products at play in the data and analytics world right now so as usual there are no straight forward, black and white solutions or 'perfect' architectures for all organisations to adhere to.  
Based on Exasol customer feedback as well as my own personal experiences at adidas over the last decade or so, one thing for sure I can recommend is to "avoid hype tools" at all costs.  From a 'storage' pov, the cloud has certainly developed and even matured since the early days of on-prem Hadoop clusters, (mostly now consigned to scrap heap of failure), to the Amazon / Microsoft / Google options we have now. But there is a reason why AWS is one of the most lucrative parts of the Amazon organisation right now - compute costs!   Majority of companies moving to the cloud have not done due diligence on the long term perspective mainly because it has been so difficult to forecast costs with any accuracy due to the 'unknown nature' of the task in hand.  
Vendor lock in is always a topic of discussion - both cloud and no-cloud - and will continue to divide for years to come.  In larger organisations there are normally huge 'political pressures' from C level and/or global architects that have their own agendas. If we were to design landscapes and architectures to cope with the every changing nature of Data and Analytics we would do well to look at tools and systems that provide the following characteristics:

RELIABILITY + MATURITY --> focus on adding business value.  Hype tools bring steep learning curves (tech debt.) so consider tech that has and will be with us long term

SIMPLICITY --> consider low code GUI tools vs  complex code

ADAPTABLITY --> still relevant in10 years time?

CREATIVITY --> open and flexible to encourage innovations

Of course - this points to tools like Exasol - but it is no coincidence that Exasol has, and continues to be, the go to solution for fast reporting and analytics at a Global Sports Brand you all know - and has seen the likes of Hadoop, SAP Hana, Oracle 12i, Oracle Exadata all come and go and has adapted from on-prem, managed service, full cloud and could easily revert back when the winds of hype/politics start to blow from a new direction 🙂