Should we move to AWS and keep ourselves cloud agnostic?

Should we move to AWS and keep ourselves cloud agnostic?

I just got back from the 2015 AWS Re:Invent conference and my brain is full of cool AWS information and ideas to help optimize our AWS usage. One of the best parts of this conference was the information exchange and experience sharing between many attendees. Learning from each other's experiences, mistakes and successes is what makes attending this conference so valuable. 

A big question many different attendees were asking themselves:  Should we move to AWS and keep ourselves cloud agnostic? 

Here are some concerns from those that think agnostic may be the way to go...

  • Worried about vendor lock-in and mentioned bad experiences around oracle fining them lots of $ after an audit… 
  • Worried about out of control costs
  • Scared about security and mentioned the Code Spaces company shutdown after AWS cloud security hack

I personally think all these are valid concerns but should be able to be resolved with better knowledge of AWS. 

Here are some comments from those trending away from an agnostic could...

  • First you will spend lots of time and energy building a layer to keep your automation agnostic 
  • You will be limiting yourself from a large set of features offered by Amazon that could help you reduce cost and increase your time to market.
  • Amazon infrastructure is rock solid. Over a period of one year, you can count customer impact caused by failures on AWS infrastructure in MINUTES. When compared to owned datacenter infrastructure, where you count customer impact minutes in DAYS. DDoS attack, Network issues, cooling issues hardware failures, sound familiar…

I personally think keeping yourself cloud agnostic is a waste of time. We should be focusing our time on improving our automation, ensuring strong security and optimizing our AWS utilization in order to reduce cost. Please share your thoughts one way or the other.

Thanks

Eric, good thoughts. I am an avid fan of AWS with my dev/ qa and prod workloads hosted in AWS. While I agree that tactically your suggestions will work fine for a enterprise, I prefer a pluralistic provider set as part of my cloud migration strategy... I enlist myself to the "Agnostic" group! Here is what drives me to that... 1. The cloud prices are still dropping. If all of us start worshipping AWS, a monopoly in this market would slow down the "race to zero" for the IaaS costs! 2. AWS have been thought leaders in this space for the last few years. Given the trends we have seen in history, these bursts of innovation/ disruption usually start to taper with organizational size. The larger they grow... ??! 3. Being part of this disruptive space means having the ability to mix and match providers and get the best. If I could run my .net shop better on Azure, I don't want a lock-in to restrict me I could go on...

Like
Reply

Eric, our company leverages AWS underneath our SaaS offering. We've made an investment in Amazon. We like the product. I've found that no matter how "open" you make things, its often hard to move off of something until it becomes painful. In the case of some of Amazon's competition, they are making it painful for people to stay. Amazon offer a lot of great building blocks for services and infrastructure making it an attractive choice. Most developers and architects want to solve problems. When scaling and capacity are a few mouse clicks away, its one less thing to worry about. I think the cloud offers more competition and new ways to solve problems by connecting solutions between cloud providers in unique ways. Already companies are creating best of breed applications by combining services from Azure, Bluemix and AWS. The inbound infrastructure model of the 90's was a true lock in. At the end of the day, we all lock in at some level. We like the idea of simply pointing our Apps at a new provider, but the more likely scenario today is to rewrite the application.

Like
Reply

If you want to be cloud agnostic, that's a choice. In making that choice, you need to ask yourself, "Where else am I locked in to a vendor?" Have you weighed the benefits of not being locked-in by that vendor as well? LIke all the decisions we make regarding our infrastructure, there are pro's and con's to choosing a vendor. Weigh those appropriately and include the potential costs of wanting/needing to move to another provider.

Like
Reply
Robb McCune

Vice President IT & Security

8y

The same logic would dictate not to leverage iRules on the F5. This scenario has been around for awhile just in different parts of the data center. Its a managed risk approach until you take on the idea to turn a blind eye to it, then its at your own peril.

Gerco Grandia

Solutions Architect in Data Analytics & Artifical Intelligence team at E.ON Digital Technology

8y

I have worked with so-called multi-cloud management platforms and was less than impressed with them. With the specific service I used you either had to use the common denominator (which was very limited) or you had to use cloud-specific configurations, the latter compromising the multi-cloud strategy. When we had to upgrade our subscription (including a significant fee raise) to use a limited version of a feature that AWS launched quite some time ago, I lost belief. From a HA perspective, the need for multi-cloud is fairly limited, as for instance the AWS regions are almost completely isolated. From a vendor lock-in perspective, you need to find the sweet spot. Enjoying the nice features and competitive advantage it might bring, vs. not being 100% tied to one vendor. I usually prefer a strategy I call 'pragmatic portability' where I try to minimise porting effort by using tools/standards such as Docker and Chef, abstracting some high level resources such as storage systems, messaging etc. but also consciously enjoying some vendor specific services which are really helpful and the alternative would be a lot of coding.

To view or add a comment, sign in

Insights from the community

Explore topics