Shulou(Shulou.com)05/31 Report--

This article is about how to analyze serverless architecture and its four major disadvantages. The editor thinks it is very practical, so I share it with you. I hope you can get something after reading this article. Let's take a look at it with the editor.

Background introduction

In 2009, the industry proposed the concept of DevOps. Wikipedia defines DevOps as "a collection of processes, methods and systems used for communication, collaboration and integration among software development, operation and maintenance, and quality assurance departments."

In 2011, Forrester issued a report "expanding DevOps to NoOps", predicting that in the near future, some enterprises will increasingly rely on the cloud, and developers will be able to more automatically carry out operation and maintenance operations such as program building (building), testing and deployment, and eventually reach NoOps.

Although the term means that these companies will no longer need operators, the report is intended to talk about developers using more automated tools that require less human intervention. Then PaaS is seen as a * way to implement NoOps.

In 2014, cloud manufacturer AWS launched a "serverless" paradigm service.

Initially, "serverless" was intended to help developers get rid of the setup and management of server devices needed to run back-end applications. The goal of this technology is not to achieve "serverless" in the true sense, but that third-party vendors are responsible for the maintenance of the back-end infrastructure, providing developers with the required functions in the form of services, such as databases, messages, and authentication. This service infrastructure can often be called a back-end as a service (Backend-as-a-Service,BaaS) or a mobile back-end as a service (MobileBackend-as-a-service,MBaaS).

But AWS Lambda, released by Amazon in 2014, takes the "serverless" paradigm to a whole new level, providing a new system architecture for applications running in the cloud. Instead of constantly running the process on the server waiting for HTTP requests or API calls, you can trigger code execution through some event mechanism, usually by running a simple function on a server in AWS. Some people call this model function as a service (Function-as-a-Service,FaaS).

Serverless architecture (also known as FaaS) means that enterprises or individuals do not need to purchase, lease, or configure physical or virtual servers to support back-end code. Serverless solutions typically consist of Web servers, Faas layer, Security token Service (STS), user authentication, and databases.

Serverless code can be used in conjunction with traditional server-style code, such as microservices. For example, we can write part of the code in a Web application in the form of micro-services, while the other part can be written in the form of serverless code. Alternatively, applications that do not require any server configuration elements in writing can be serverless.

FaaS provides a platform that allows developers to execute code in response to events without having to build and maintain a complex infrastructure, only through third-party applications or services to manage server-side logic and state.

Four drawbacks of serverless computing

1. Problems caused by third-party API systems

Vendor control, multi-tenancy issues, vendor locking, and security flaws can all be caused by third-party API. Abandoning system control during the implementation of API may lead to system downtime, forced API upgrade, lack of functionality, unexpected restrictions, and cost changes. In addition, the multi-tenancy problem also exists in other cloud computing frameworks.

Salesforce (PaaS) imposes some regulatory restrictions because of its multi-tenant cloud structure, and developers must avoid related problems as much as possible when using Salesforce. Specifically, multi-tenant solutions tend to have problems in terms of security, stability, and performance.

two。 Missing operating tools

Developers rely on suppliers to provide debugging and monitoring tools. Generally speaking, the task of debugging a distributed system is very difficult, and it is usually necessary to access a large number of relevant indicators to determine the root cause of the problem.

3. The complexity of the architecture

Developers often spend a lot of time evaluating, implementing, and testing specific features before finally deciding how they should be subdivided. The number of functions involved in an application call operation should be balanced. Managing too many functions will undoubtedly complicate the problem, while ignoring granularity will eventually turn the micro-service architecture into a "mini-overall" architecture.

Currently, Lambda, a computing service provided by AWS, which can execute code in a massively parallel manner in response to events, has limited the total number of concurrent executions that users can run on all lambda expressions. The problem is that this restriction applies to the overall AWS account. Some organizations use the same AWS account for production and testing. This means that if a staff member in the organization embarks on a new load test and tries to execute 1000 concurrent Lambda functions, the production application will immediately encounter a denial of service (DoS) condition.

4. The difficulty of implementation

Integrating and testing serverless applications is very difficult. Compared with other architectures, the integration unit of serverless FaaS (that is, each function) is much smaller, so we need to integrate a large number of units to complete the test properly. In addition, there are some issues related to deployment, versioning, and packaging. You may need to deploy a separate FaaS component for each function in the overall logical application. This means that you cannot uniformly deploy a set of functions in an atomic way, and because there is no concept of versioned applications, atomic rollback (atomic rollback) cannot be implemented. In this case, you may need to turn off any event source that triggers the appropriate feature, deploy the overall feature group, and then restart the event source.

The above is how to analyze the serverless architecture and its four major drawbacks. The editor believes that there are some knowledge points that we may see or use in our daily work. I hope you can learn more from this article. For more details, please follow the industry information channel.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.