Four basic strategies for migrating cloud computing workloads

actually , It's easier said than done .IT Engineers don't suddenly want to move the entire application from one cloud platform to another , This is not what the term cloud migration really means . On the contrary , Workload portability reflects that both short-term choices and long-term flexibility are possible . Decide to run a specific application in a specific cloud platform or on premises environment , It does not mean that you need to run in this running environment all the time .

Containerization of modern software development and operation 、 Preparation and other aspects are the basis of the operating environment . Of course , Workload portability also indicates , The enterprise has at least two cloud platforms or internal deployments / Bare metal environments to choose from .

Migrating cloud workloads may encounter several important problems .Red Hat Company technical evangelist Gordon Haff Express , Workload portability is an important term for hybrid cloud and edge architectures .

Migrating cloud computing workloads 4 Methods

Here are four different ways to migrate cloud computing workloads , To help users more carefully understand how to run and migrate workloads in multiple operating environments , And develop an overall hybrid cloud or multi cloud strategy .

(1) Set standards for migrating cloud computing workloads

Many mixed clouds and cloudy environments begin in a temporary or even accidental way . It's natural , But as Haff Noted , They should eventually be replaced by a more purposeful strategy .

First , There should be clear criteria for running and migrating workloads in a given environment .

Faction Chief technology officer of the company Matt Wallace say ：“ There are many reasons to decide where to run the workload . The hardest part is , Because the workload of the team or partner is located in different cloud platforms , Or you need to access different services , So there is no right answer .”

therefore , Focus on the specific reasons that are important to the enterprise , Let them guide the choice of enterprises .Wallace Shared several examples ：

• Access to other applications and data —— Also known as “ Data gravity ”, When performance / When delay becomes a major problem , This is usually a driver .
• Work with other teams and partners —— If they use a specific cloud platform , You could do the same thing .
• Toolsets available in specific cloud platforms —— They are not all the same .
• Geography / Location problem .
• cost .
• scale —— For example, predictable 、 The difference between a stable workload and a workload that may grow or surge in resource demand .

There are some standards , Additional specificity in the goals or needs of the enterprise will be beneficial . Its performance is quite extensive , For example, the balance of performance is delay . Define the actual meaning of these terms for the enterprise and Its Applications , A more refined decision matrix will be provided , To match the workload to the right environment .

Again , The choice of cloud platform is not one size fits all , Especially when going beyond the core infrastructure .

Liberty Mutual Insurance Senior architect of the company EricDrobisewski say ,“ In any cloud platform environment , Infrastructure services are all bets on the desktop . In addition to these core services , Also determine the key elements of public cloud providers , These elements can bring differentiated value to the business of the enterprise , And seek to use these elements to bring greater value faster .”

(2) Make sure everyone and everything work well

as time goes on , Hybrid and multicloud environments tend to become more distributed and diverse . A key to effectively managing and migrating workloads is the ability to make changes without destroying everything , It should be possible to add new tools or services , And make it compatible with the existing technology stack .

Wallace This strategy is refined into a term here ：“ framework ”.

He added ：“ Design architecture avoids getting caught up in interpretation . It is useful to take advantage of abstract tools that provide portability or consistency . Use centralized identity and security assertion markup language (SAML) Authentication is also useful for standardizing other things .”

in fact , Standardization is a major advantage in the integration strategy . Open standards are even better , Especially considering the changing speed of Cloud Computing .Drobisewski Pointed out that , This is beneficial for both initial integration costs and long-term flexibility .

Drobisewski say :“ If possible , Using open specifications and standards adopted by various cloud computing providers will help simplify integration , And improve interoperability .”

SAS Senior Software Development Manager of the company Justin Dempsey Express , Don't know “ everything ” What do you mean by , It's hard to keep everything in harmony . His team found out , Creating a matrix of tools and applications across multiple cloud platforms is very useful . This can help everything from identifying gaps to ensuring the security of the software supply chain . It can also inform decisions about workload portability .

Dempsey say ：“ Enterprises create a matrix of tools for their control , And point out which tools are irrelevant to Cloud Computing 、 Non cloud portable or cloud specific , This helps enterprises assess the risks involved in moving from one cloud platform to another , Or create an architecture that needs to span cloud computing providers .”

Managing as much code as possible is another important strategy here .

Dempsey say :“ Face ‘ Everything is code ’ The direction of work is to promote consistent delivery 、 Methods of complying with governance controls and mandatory testing standards , These standards can ensure the harmony between the new operating environment and the existing operating environment .”

(3) Manage and optimize costs

Cloud computing costs are often oversimplified as absolute and extreme costs , for example “ Using cloud computing services is cheaper !”( It's not always true ) or “ Why is my cloud computing bill so high ?”( There may be many reasons .)

This is another area that requires careful design and Planning .Faction The company's Wallace Pointed out that , Many of the contents that may be classified as infrastructure costs are actually application level problems .

Wallace say ：“ If we build a three-tier automatic expansion architecture in Cloud Computing , use API Gateway and serverless capabilities to handle things that can be done at a fraction of the cost , Then it will pay a huge price for using cloud computing services .”

just as Red Hat The company's Haff As I said before , The cost of adopting cloud computing services will indeed become more expensive . This does not mean that they should not be used ,Haff explains ,“ But we need to know where they provide higher value for enterprises , And where you should consider running workloads in on premises facilities .”

A comprehensive understanding of costs is essential , Especially when making informed decisions about running and migrating workloads and data .Wallace Use deep cold storage as another example , Therefore, the initial cost may not seem expensive .

Wallace say ：“ In a cloud platform , The cost of retrieving data from the cloud exceeds 4 Annual storage costs . This is not a problem for cloud computing providers , It's the user's response to ‘ Store and forget ’ The huge demand for use cases , These use cases can replace tapes in offline repositories , But if users don't match services to use cases , Then it will face high costs .”

In terms of workload portability and cloud computing costs , There are two main areas that need attention ：

• visibility ： Effectively managing cloud costs can be attributed to the ability to answer the following questions ：“ Who is using what ?”, Optimize cloud cost to be able to answer “ Why? ?”,Wallace The use of financial terminology clearly raises this question ：“ Who spends how much on what services ?”, If this is a black box , Then it will be difficult to achieve the cost target .
• Data flow ： Migrating cloud computing workloads may generate additional ( Sometimes it's accidental ) cost , These costs are related to the data flow in and out of the environment —— It is usually called data input and data export .

Both methods require payment , But the cost of data export usually needs attention .

SAS The company's Dempsey Express ：“ Data export costs may increase rapidly , Especially for data migration across multiple cloud platforms or cloud computing regions .”

Wallace The example of deep cold storage mentioned is one of many possible scenarios involving unexpected cloud computing bills caused by data export fees .

Wallace say ：“ This is the most significant in terms of network traffic , For example, someone opens a network gateway in the public cloud to connect to the virtual network , You may pay for using the gateway every day 2.4 dollar , But as an extreme example , Every day can produce 10800 Data transmission cost in US dollars .”

When migrating workloads between cloud platforms , The possibility of soaring costs will increase .

Wallace say ：“ When it comes to cloudy , The risk is magnified , Because network traffic outside cloud computing is more likely to incur higher costs . This is a generalization , But it should be noted that , These data flows must be understood .”

(4) Developers need to keep it simple and fast

Last , Don't forget developers . Developer experience is everything now .

As mixed and cloudy environments become more diverse and complex , Have some advantages ( Including the control and flexibility to match the workload with the best environment according to the standards determined by the enterprise ) It depends on preventing unnecessary friction in the development team .

Wallace Pointed out that , It actually depends on many different factors , For example, developers 、 Application Portfolio 、 The code base 、 Your mission, etc .

Wallace Express , This is an ideal scene , Combine the benefits of multicloud with the rich experience of developers , It may be a serverless model , Developers can develop in an on premises or cloud computing development environment , There is little infrastructure to maintain , And in API The gateway and other components have built-in tools such as throttling restrictions , To avoid out of control cost caused by out of control code in development .

Will write 、 Excellent tools that minimize friction between testing and deploying code are good for business and developers , At the same time, it is also the basis for achieving real workload portability .

Wallace say :“ This design pattern is important for maximizing any cloud platform 、 The portability between on premises data centers and edge deployments is also very good .”

Drobisewski Pointed out that , The benefits of hybrid cloud and multi cloud ecosystems may allow developers to simplify them .

Drobisewski say ：“ Invest in a single market that unifies technical support and plans a series of well structured models , These models are both safe and cost optimized , Will accelerate developer support , At the same time, cultivate a culture of reuse .”

Last , If the portability of cloud computing workloads is a priority , So mixed clouds and cloudy clouds can actually work well together , They can be interdependent , So as to improve the development speed of developers .Dempsey It is recommended to avoid focusing too much on specific methods or project management styles .

Same as the cost , Developers' development speed should also be reviewed at the application level .

Dempsey say ,“ Which application in the enterprise does not provide some kind of abstraction —— What aspects of the application stack are closely tied to a particular technology or vendor ? They may be the source of friction , If the goal is to decouple and focus on creating robust data delivery pipelines , This will provide long-term flexibility and integration opportunities for developers and data consumers .”