Cloud migrations are important elements of digital transformation for businesses.
Despite optimistic expectations about the cloud eating up the world, there still are a lot of enterprise workloads that run on local, on premise server rooms or with hardware hosting companies.
That was clearly visible in the survey Avenga did with CIOs.
Admitting the pace of cloud transformation is slower than anticipated also means there’s time to prepare businesses better and choose the optimal paths and patterns, learning from the lessons that others experienced.
Let’s discuss the strategy called Lift and Shift which happened to be recommended just a few years ago; and now it’s time to call it an antipattern. What is it? Why isn’t it considered the best option anymore? Let us help you find out.
Lift and shift is the approach of moving virtual machines from a local infrastructure to a cloud infrastructure while avoiding changes as much as possible in the machines and in the network configurations.
No architecture changes are required and new hardware will be emulated as closely to the old ones so as to maintain a very high degree of compatibility with the on premise infrastructure.
→ Explore Evolutionary architectures in practice
Cloud vendors are waving their hands and sharing success stories, to lure clients into moving their infrastructures to the cloud – server by server.
It’s the simplest cloud migration possible. The closest to copy & paste as it can get. There’s no doubt about it.
Advanced migration tools from the local VMW are ESXi environments for AWS, Azure, Google, and others that help with this process.
It’s been working quite a long time, so the tools and processes have been proven in action, and the probability of a successful migration is high.
There’s never ‘just. . . ’, as there’s effort and planning required to do it, but it definitely works.
“If it works – don’t fix it.” Sometimes this is temporarily good, but in the longer term it’s a curse. Avoiding change in IT or business is simply delaying the change.
→ Read more about Legacy migration by a thousand features
Imagine someone who just bought a new house or apartment. The next step is to move there.
Most of us prefer to use this as a motivator to get rid of old stuff, buy new things, and ‘configure’ our next living place into a different and improved place.
Of course there are also people who move everything and want to keep things the same as they were before, thus missing their chance for optimization, however, they can physically move to the new place faster.
Cloud vendors are always encouraging CIOs to do the cloud lift and shift. One of the reasons is just to have the client on their side in the cloud. Then later think about how to optimize the infrastructure and how to really benefit from it.
The prevailing opinion is that when servers are kept up, as they were in the local infrastructure, the cost will be much higher. But with smart turn on/off schedules, the cost may be even lower than in local infrastructures.
However, unexpectedly it seems that the systems are used at night for data synchronization, as sources of data for the reporting systems, and . . . cannot be just turned off even though it seemed so easy based on working hours.
Additionally, front-end systems for sales and self-service cannot be shut down either as they need to be constantly available to clients and business partners.
Database systems optimized for the cloud are usually modified or are entirely different engines than local on premise database servers, which take tens of gigabytes of RAM and tens of CPU cores to be prepared for a traffic spike. And when these spikes come, it’s very often not enough.
The cost will almost always be higher than planned and almost always be higher than in the local infrastructure, and 100% of the CIOs I know admit that. This is simply not a cost effective approach.
After a simple lift and shift, the servers are the same and configurations are the same (well almost).
One of the key promises and motivations for moving to the cloud is scalability.
Business examples: e-commerce is experiencing major changes in customer traffic especially on Black Fridays, during Holiday Seasons, and during major promotions of the products that result in severe performance issues resulting in timeouts and discomfort for the clients.
Many of us experience humiliating messages on our customer’s screens such as “our servers are overloaded” or something similar. That’s the most visible manifestation of the scalability problem, which directly hurts the business (less transactions) as well as indirectly (image of a company unable to cope with the new digital reality).
Cloud migration should be deeper and involve refactoring of application’s architecture to enable scalability. Sometimes it’s near impossible to do with old terminal applications, but in most of the cases typical three tier architecture can be at least cloud enabled to benefit from multiple nodes of scalability.
Changing the architecture from monolithic to fully microservices may not be a viable option, but at least a form of modularization is usually possible and makes sense from both maintenance and application evolution standpoints.
Lift and shift won’t make your apps more modular and as a result containerized. So the opportunity to prepare apps for the containerized world will be possible later, which in busy IT terminology often means never.
Serverless – impossible
There are scenarios where a serverless paradigm is very helpful, but it requires deep cloud migration and lift and shift won’t make your business logic accessible as cloud functions automatically.
If anyone ever dreamed of NoOps helping here, that won’t be possible either.
The servers and services are the same, so the operations team and processes are usually not changed at all. Therefore, you’ve just moved to the new apartment and all the things you loved and hated are simply in a new location.
In our discussions with CIOs the prevailing trend was to embrace hybrid clouds and not to migrate applications which are close to their end of life.
Cloud IS NOT just someone else’s computer. The term was coined by the cloud fearmongers and is really hurting the benefits of the cloud.
It’s the entire work that’s being done in the background by the cloud teams, their unique knowledge and experience, that you buy.
Cloud lift and shift is now considered as the last resort option. I can imagine the case where a local infrastructure would require major investments after recent events such as major hardware failure.
Lift and shift made the IT community emphasize the difference between a shallow copy & paste migration and a deeper cloud transformation.
Cloud transformation is a crucial part of your ongoing and neverending digital transformation. It’s an enabler for scalability, flexibility, new architecture (i.e., serverless) and organizational paradigms (i.e., NoOps). This is the prevailing mindset of the most innovative CIOs and CDOs.
Avenga with its vast cloud experience can definitely help you do it in a way that will maximize the return on your investment.