Cloud migration: paths to failure and success
19 November 2019 | 0
For most companies, migrating to the cloud is no longer a question of if but when. By moving applications to the cloud, you can improve security, data access, scalability, and IT flexibility, just for starters. Moving to the cloud can also save you money.
However, be forewarned: Not all cloud deployments go smoothly. Migrations often take longer than expected, or they fail completely, resulting in wasted time and expense. It is not unusual to discover, after moving an app into the cloud, that it does not work as well there as it did on-premises.
The result might just be another migration – back to the data centre.
A recent study, sponsored by security provider Fortinet and conducted by supply chain specialists IHS Markit, found that many companies, 74% of those surveyed, have moved a cloud-based app back on-premises after failing to achieve the anticipated benefits.
This is not a new problem. A Google search of cloud migration failures will find examples going back several years. We have been discussing the problem for some time, and the problem is not a failure of technology, but a failure of leadership.
Here are five leading causes of cloud migration failures, and what you can do to succeed.
Cloud migration failure #1: Lack of a good partner
The first step is to realise you cannot do this alone, especially in the beginning. You will need a partner, whether it is a global professional services company like Accenture or a local consultancy. And that is a decision that should be made with careful deliberation and some outside input. Ideally, you have a network of peers in your industry and geography who can help you pick the right consultant for the job.
“Choose your partner carefully. Get references. You need a very reference-able partner, one who can step you through the process who has not only tech capabilities but change management capabilities as well,” said Joshua Greenbaum, president of Enterprise Application Consulting.
A good cloud migration specialist can help you identify the best applications to move, determine how to integrate legacy systems and cloud services, and plan and execute the migration. A good partner can also help you work out an effective hybrid or multi-cloud strategy.
Cloud migration failure #2: Failure to adapt to the cloud
One of the most common mistakes companies make is letting their apps run in the cloud the same way it did on premises. That, says Tim Crawford, president of Avoa, a CIO consultancy, is a huge and common mistake.
“On-prem apps are used to consuming resources at peak,” Crawford said. “The cloud is designed to use resources when you need it and give them back when you do not. But the traditional app is not built with the level of autonomy and orchestration to take advantage of cloud.”
Too many customers forget that every single bit they run on a public cloud is metered and they will have to pay for it. They let unmodified apps run at full tilt, eating up compute cycles, and the bill comes a month later. Simply lifting and shifting an app to the cloud is a recipe for sticker shock, at the very least. At worst, you are facing a return to on-premises.
Cloud migration failure #3: Not having the right skills in-house
If you think you can manage your public cloud or even hybrid cloud with the old skills and approaches – ITIL framework, waterfall processes, monolithic applications, operational silos, etc. – you are in for a rude surprise.
You need skills to manage dynamic infrastructure, containers, automation, microservices, and so on. Problem is, so does everyone else! New technology will help, but attracting, training, and retaining skilled talent is still critical.
“A cloud operating model moves IT away from traditional, static, monolithic software management using standalone on-premises legacy tools and suites, to an environment of highly distributed, dynamic, atomised, and abstracted services managed with multiple often cloud-based point solutions,” said Andi Mann, chief technology advocate at analytics firm Splunk. “IT needs new skills for managing the cloud platforms themselves and also the containers, microservices, APIs, SaaS systems, and so on.”
Cloud migration failure #4: Not bringing in stakeholders
Projects need very good governance, and that means involving everyone touched by the transformation to the cloud. Oftentimes projects are driven by the IT department and then they tell those impacted after the project is done.
“This is way more common than anyone would admit,” noted Greenbaum. “A lot of this is basic project management, checking that the right people are in the steering committee and getting the right info. Often they are not invited in until it’s too late.”
One example Greenbaum recently encountered was a company that made a significant change to the customer experience in the process of moving to the cloud. Unfortunately, it did not take into account the impact on the supply chain. As a result, the supply chain team was not part of the sales overhaul. Only after the migration was done did those in the supply chain group realise what had happened and found out they could not meet the new demand generated by the revamp.
Cloud migration failure #5: Unrealistic expectations
A move to the cloud can deliver great benefits – speed, agility, cost reduction, strategic focus, scalability, reach, and more – but it can introduce legitimate risks, too. Gaining maximum advantage from a cloud deployment starts with avoiding the lure of the hype cycle and having realistic expectations both for what you can achieve, and for potential new risks.
Leadership especially expects to save money with cloud but that is not always the case, especially if you commit failure #2 on this list and do not rearchitect your application. Cloud adopters often also expect to be able to do a lot less work in adjacent areas, but cloud infrastructure only replaces the servers, not your IT people.
Do not think you can move to the cloud and get rid of all of your DBAs, security ops, service desk engineers, and other software experts. And if you are running a hybrid cloud like the overwhelming majority of enterprises, you will still need hardware support for the physical assets you will be holding onto.
Cloud migration steps to success #1: Bring the A team and demand the same of your partner
Demand for cloud services is outstripping supply and there is a dearth of highly qualified people to do the work that needs to be done. Greenbaum said he has seen many projects get hung up due to a lack of personnel.
“Projects succeed because the customer brings their A team to the grid and demands their systems integrator bring its A team as well,” Greenbaum said. “If you don’t put your best people on it you risk having an inferior outcome.
Cloud migration steps to success #2: Be picky about what goes to the cloud
A common mistake companies still make is moving everything to the cloud, when not everything belongs there. Crawford said to put the most standard line of business apps in the cloud and keep the unique code in-house.
“If it’s not differentiating for your business, consider moving that, like email and calendaring, ERP, HCM,” Crawford advised. “Core back office functions are critical and needed but is that where your IP differentiates you from your competitor? No. Those are good opportunities to move to the cloud,” he said.
Cloud migration steps to success #3: Innovate and differentiate
Since you should be refactoring the apps for the cloud anyway, consider it your opportunity to embrace new methodologies and designs. Rearchitect as many on-premises apps as possible for a cloud native design, where the app is elastic and scales up and down as needed. Containerise your app, so it runs on Docker and is managed by Kubernetes. All of the major cloud providers offer services to assist with Kubernetes both on-premises and in the cloud.
“The most successful organisations I know have used the fundamentally different nature of cloud to innovate, not just replicate – delivering that new prototype they never could before, getting service to a level that customers never expected, and engaging in new ways with new applications for new markets,” said Mann.
Cloud migration steps to success #4: Have a cohesive strategy
Approaching the cloud strategically means rethinking budgets, organisation, processes, skills, security, data integration, and on and on. Technology is only a small part, but one where a cohesive strategy can quickly unravel. A successful migration includes making conscious portfolio decisions of what to keep on-prem and what to move, which platforms to stick with or abandon, and how to refactor applications to take advantage of the cloud’s benefits. By standardising on common compute, storage, and database platforms, you can reduce both complexity and the costs of management and operations.
Keeping things simple also means avoiding an overcomplex migration and biting off more than you can chew. Bad things happen when the scope of the project is too vast or the time frame or budget is too small. Do not do it all at once. Break up the project in stages and tackle them one at a time. Adopt an iterative, devops-like approach. Do one piece, make sure it works, then move on to the next piece of the project.
Cloud migration steps to success #5: Consider a new data model
A migration to the cloud could mean the opportunity for a completely new data model. Putting data in the cloud is an opportunity to expand the data model to one that is much wider. For example, moving to a more customer-centric model could mean bringing in more data from many different sources.
Your old on-prem data might have a simple customer input, like name and address, but your new cloud data might draw from social media, IoT devices, and other sources. Or you might even migrate to a completely different data analysis platform. Amazon Redshift is PostgreSQL-compatible, but Google’s BigQuery uses different types than typical SQL or PostgreSQL. And Snowflake supports various formats for semi-structured data.
“In theory, you are changing business practices, so what you need from your data is different,” said Greenbaum. “Changes in data quality are as political a decision as anything else. It’s not just a simple question of let’s put our data in the cloud, it becomes a real change management problem.”
IDG News Service