Should enterprises choose server free computing?

Until recently , Almost every product manager will treat him / Her engineering resources , Divided into two independent teams —— Development team and operation and maintenance team . Development teams are usually involved in coding 、 Test and build application functions , The operation and maintenance team is responsible for the delivery of applications 、 Deployment and operation maintenance .

When the development team builds e-commerce applications , The operation and maintenance team will set up a server to host the application . Building a server involves many aspects , These include ：

• Choose the right hardware and operating system
• Apply the required patch set
• Set up the required server environment , Such as JDK、Python、Tomcat、NodeJS etc.
• Deploy 、 Configure and provide practical applications
• Open and secure the appropriate port
• Build the required database engine
• …… The list continues .

besides , Managers also have a headache about Capacity Planning . After all , Any important application should always be maintained 100% You can use 、 Reliable and scalable . This requires the best investment in hardware . as everyone knows , In some critical periods , Hardware shortages can lead to business losses , And hardware redundancy will damage profits . therefore , Whether the application is for local data centers , Or for cloud infrastructure , Capacity planning is crucial . up to now , Obviously , Enterprises not only invest a lot of energy in function construction , Also spent a lot of time on function delivery .

Serverless computation Serverless computing It aims to provide a seamless way to deliver functions , There is no need to worry about the setup and maintenance of the server . let me put it another way , Serverless computing platform provides a “ Ready to use ready-to-use” Environmental Science , Enterprises can build and deploy applications into smaller functions as soon as possible . That's why this method is called “ Function as a service Function as a Service”（FaaS）.

please remember , There are still servers in serverless Computing , But it was created by AWS、 Microsoft, Google, etc FaaS The supplier is responsible for .

for example ,AWS With “Lambda function ” Provides a serverless computing environment in the form of . Developers can choose to build the application into a set Lambda function , These functions can be used NodeJS、Java、Python And some other languages .AWS Provides an off the shelf environment to deploy these functions . It also provides a ready to use database server 、 File server 、 Application gateway and authentication server .

Again , Microsoft Azure It also provides an environment , It can be used C# And other languages Azure function .

Why choose no server ？

There are two main factors driving the popularity of serverless Computing .

1、 Ready to use environment

obviously , This is the biggest selling point of serverless Computing . Enterprises do not need to purchase in advance / Subscribe to hardware or instances , And don't worry about licenses , And setting up and configuring the server . They don't need to worry about expanding and shrinking . All this is done by FaaS The supplier is responsible for .

2、 Optimal cost

because FaaS Suppliers always charge customers based on the utilization rate of the environment （ Pay per use mode ）, Therefore, enterprises need not worry about early-stage costs and waste of resources . for example ,AWS according to Lambda Number of requests received by function 、 The number of queries running on the data table and other indicators to charge the client .

The challenge of server free computing

Like any other method , Server free computing is not a perfect method that everyone can blindly follow . It also has a series of limitations . Here are some of them .

1、 Supplier lock up

When using serverless Computing , The first and most important problem is ,Lambda or Azure Functions such as... Will be provided by the supplier API To write the . for example , Use AWS Lambda API The function written cannot be deployed to Google Cloud in , vice versa . therefore , Serverless computing forces enterprises to , Only one supplier can be used . also , The success or failure of an application depends not only on its function , It also depends on the ability of the supplier in terms of performance .

2、 programing language

No serverless computing platform supports all programming languages . Besides , For the programming languages it supports , It may not support all versions . thus , The application development team can only choose the language provided by the supplier . In terms of team capabilities , This may be very critical .

3、 Optimal cost , Really? ？

Not necessarily , It all depends on the use of resources . If your application is under a huge load , For example, millions of requests per second , Then you may pay too much . On such a scale , It may be cheaper to have your own server locally or in the cloud . This does not mean having Web Large scale applications are not suitable for server free computing . in the final analysis , It still depends on how your platform is built , And the agreement you signed with the supplier .

4、 The ecological system

No application is written for an isolated environment . It always needs other components , Like data storage 、 database 、 Security engine 、 gateway 、 Message server 、 queue 、 Cache, etc . Each platform provides its own set of such tools . for example ,AWS Provides Dynamo DB As its NoSQL One of the solutions . obviously , Other suppliers also provide their own NoSQL Solution . therefore , The team will be forced to build the application based on the selected platform . Although most businesses FaaS Suppliers provide multiple components for specific requirements , But not every component can be the best of its kind .

Why not consider containers ？

In the past decade , Many of us have migrated to the containerized deployment model , Because they provide a lightweight alternative to expensive physical or virtual machines . With Kubernetes After the editing tool , We are happy to deploy containerized applications , At the same time, I am satisfied with Web Scale requirements . Containers provide a certain degree of isolation from the underlying environment , This makes deployment relatively easy . however , We still need hardware （ Local or cloud ）、 license 、 The Internet 、 Investment in configuration , This requires forward-looking planning 、 Appropriate technical capabilities and careful monitoring . Serverless computation , Although it also has its own advantages and disadvantages , But it frees us from these responsibilities .

Looking forward to the future

We are in the process of continuous development 、 The era of continuous integration and continuous deployment . Every enterprise is facing competition . Time to market Time to market（TTM） Attracting customers 、 Two aspects of retaining customers , Play an important role . In this context , Enterprises like to spend more time to launch functions as soon as possible , Rather than struggling with the details of deployment and maintenance . Serverless computing has the potential to meet these needs . Big players are investing huge amounts of money , In order to make FaaS As seamless and economical as possible . The future of serverless computing looks bright .