Steps to be taken for successful migration to the cloud

Investigation shows , During the continuous spread of the COVID-19 , The adoption rate of enterprise cloud computing has risen dramatically , Now it is a rule, not an exception . in fact , according to O'Reilly A survey report recently released by the company , at present 10 There are 9 Companies use cloud computing services in some way .

Although many industry organizations' digital transformation plans are well underway , But the global spread of the COVID-19 has introduced two new factors , Force almost all enterprises to transfer their operations to the network . First , This is the main way for them to communicate and contact with customers . In the case of travel bans and physical store closures , Customers have to rely almost entirely on digital services for shopping 、 Get support 、 Participate in personalized experience and interaction with enterprises .

secondly , The full shift to remote work makes it impractical to continue using on premises hardware and computing resources . To ensure that teams working remotely can work together effectively , Moving to a cloud platform is the only option for many companies . Although the current use of statistics attests to the success of the private sector in this regard , But most companies have encountered some obstacles in their journey to cloud computing .

Obstacles to cloud adoption

Now , There are several different types of cloud platforms and multiple cloud service models . For the sake of simplicity , Tends to consider cloud computing resources from the perspective of two components ： Back end and front end . The back end is the infrastructure layer . Outside the physical servers and data centers included in each cloud computing provider , The infrastructure layer contains everything related to the information architecture , Including data access and security 、 Data storage system 、 Computing resources 、 Availability and service level agreements . The front end is the presentation layer or application program interface , Include end user profiles 、 Authentication 、 to grant authorization 、 Use cases 、 User experience 、 Developer experience 、 Workflow, etc .

not long ago , Enterprises often migrate to the cloud platform in a long period of time , According to the experience of industry experts cooperating with enterprise customers , Spend a lot of time designing and implementing the back end , Then do the same for the front end . The COVID-19 has changed that . In the past, it was a gradual process , It's a fast job with a tight schedule , Front end and back-end systems are often implemented together , End users are introduced earlier to participate in more frequent iterations .

Besides , COVID-19 was introduced and constructed 、 Consideration of the costs associated with maintaining and operating these front-end and back-end systems . Enterprises are seeking more cost savings as much as possible , Although in the long run , Cloud migration can reduce total cost of ownership , But it does require upfront investment . For those facing potential labor and financial constraints , Cost is an important factor to consider .

Aggressive schedule and cost considerations are not obstacles in themselves , But they certainly pose challenges during cloud computing deployment . So what are the other obstacles to successful cloud Integration ?

(1) Try “ Ascension and transfer ” framework

When trying to meet the cloud migration deadline , Enterprises often tend to configure their cloud resources as exact copies of their on premises settings , Regardless of cloud native services that can offset significant maintenance or performance overhead . Without considering how to use cloud native services and redesign the different components of its workflow , Enterprises will eventually bring all inefficiencies to the cloud . On the contrary , Enterprises should see cloud migration as an opportunity to consider better architecture , To save money 、 Improve performance , And bring a better experience to end users .

(2) Focus on infrastructure rather than user needs

When data leaders decide to move their business to the cloud , They tend to pay more attention to the features and functions of various cloud services , Without considering the workflow of data scientists and data engineers . Data leaders typically do not optimize for developer productivity and rapid iterations , Instead, it focuses on developing powerful and scalable back-end systems . Besides , Data professionals want to make the cloud architecture more perfect before bringing users into the cloud computing environment . But the longer the cloud computing environment is not tested by end users , The less useful it is for them . It is recommended to introduce a minimum amount of data into the initial cloud environment 、 Development environment and automation tools , Then bring in users and iterate according to their needs .

(3) Unable to access production data in the cloud

Data professionals often enable many different cloud native services to help users perform distributed computing 、 Building and storing container images 、 Create data pipes, etc . However , Before some or all of the production data of the enterprise is available in the cloud computing environment , It won't work immediately . Enterprise leaders should work with their data engineers and data science teams , Determine which subsets of data they can access in the cloud 、 Migrating data , And let them experience the benefits of cloud services . otherwise , Business leaders may find , Due to the data gravity , Almost all production workloads remain in on premises facilities .

Smoother cloud migration

Despite the obstacles , But data leaders can take many steps to ensure that their cloud computing deployment is as smooth as possible . Besides , Taking these steps will help maximize the long-term return on investment in cloud adoption ：

(1) Centralize new data and computing resources

Many enterprises provide too much or too little computing and data analysis resources , Eventually, the solution is fragmented , And insufficient documentation . therefore , Enterprise adoption is slow , Most of the user's work is done on the isolated island facilities or laptops , The entry of new data engineers and data scientists can be a messy process . Enterprise leaders can focus on the core data sets and computing requirements of the most common use cases and workflows , And focus on solutions to avoid this . Concentrating resources does not solve all problems , But it will enable enterprises to focus on the main challenges and bottlenecks , And provide support for most people .

(2) Involve users as early as possible

Usually , Before informing users that the cloud computing environment is ready for use , It takes months or even years of infrastructure management and deployment . Unfortunately , This usually leads to low utilization of cloud computing environment . To overcome this waste of resources , Data leaders should target the end user experience 、 Workflow and use case design ; Join the end user as soon as possible in the process ; Then update the iteration with them , And address the biggest challenges in order of priority . They should avoid delaying use in the name of designing a perfect architecture or an ideal workflow . On the contrary , Business leaders can engage key stakeholders and representative users as early as possible , To get real feedback on what needs to be improved .

(3) Focus on Workflow first

The enterprise should determine the core data set ( Or subset ) And the smallest possible toolset , So that data engineers and data scientists can complete 80% The job of , Instead of building a completely robust at the first iteration 、 Scalable and redundant systems . then , They can gradually gather feedback and determine the next set of solutions , Shorten the feedback loop as effectively as possible in each iteration . If an enterprise handles production datasets and workloads , So in terms of security 、 performance 、 Scalability or other functionality at acceptable and standard levels , No shortcuts should be taken . Data leaders can buy off the shelf solutions , You can also work with others to provide solutions , To bridge the capability gap .

No turning back

Cloud computing technology used to be a differentiated technology , Now it is a mainstream technology . The only way for an enterprise to gain a competitive advantage is to equip its data team with the tools they need , To better complete the work . Even the most expensive 、 The most secure and scalable solutions will not be used , Unless it is truly licensed to the end user .