Clever Geek Handbook

KOST: what is included in the new technology stack for developing cloud applications

We talk about its components and prospects.




/ photo Gozha Net Unsplash



Why a new stack appeared



There are many stacks for software development. One of the most famous is LAMP (Linux, Apache HTTP server, MySQL, PHP). The Wikipedia website and several other projects of the Wikimedia Foundation are built on it . LAMP was even used on Facebook to solve a number of problems.



But it is believed that LAMP, although it is a proven stack, is not well suited for building large, scalable applications. Therefore, today, when companies transfer workloads to the cloud , alternative technology sets are increasingly being used.



One of the most popular stacks for the cloud is MEAN - MongoDB, Express.js, Angular and Node.js (often nginx is added to the latter, which is used as a reverse proxy server). It is “tied” to JavaScript, and in this language you can write both the server and client parts. This stack is used by Paypal, Netflix and The New York Times. It is also used in the development of mobile applications. MERN and MEVN are also considered popular technology sets, in which Angular is replaced by React or Vue.js, respectively.



An alternative stack, KOST, has recently begun to gain momentum. It includes Kubernetes, Operator, Spark and TiDB. Its goal is to optimize data processing in private, public and hybrid clouds.



Components



Kubernetes . Perhaps one of the most famous tools for managing container applications. It became part of KOST because it is a kind of standard in the IT industry. He is supported by members of the Cloud Native Computing Foundation ( CNCF ). Someone even calls Kubernetes "an operating system for the cloud." Many large IaaS providers and developers of virtualization tools already offer their solutions based on Kubernetes - for example, OpenShift, VMware PKS or Rancher.



Operator . An open set of tools for managing “operators” - Kubernetes-based applications. These applications help implement additional business logic, for example, to work with third-party resources. Operator was released in 2016 by engineers from CoreOS - today Red Hat is responsible for it. The tool itself, like Kubernetes, is gradually becoming the standard for developing cloud applications. For example, several DBMS developers have already implemented it. And the number of users is steadily increasing.



Spark Another well-known open source framework for working with unstructured data. Spark uses specialized primitives for recursive processing in RAM. This approach allows you to increase system performance on some tasks ( page 1 ). The tool found a place in the KOST stack, as it is actively used to deploy large-scale cloud infrastructure, for example, in Netflix, Oracle, Visa and eBay.





/ photo by Stephanie McCabe Unsplash



TiDB . This is a distributed DBMS that combines transactional and analytical data processing (Hybrid Transactional / Analytical Processing, HTAP). It scales horizontally, has high availability and follows the principles of ACID : atomicity, consistency, isolation and durability. The system was originally created with the aim of working in the cloud to simplify the deployment and support of applications. This is one of the main reasons why she was chosen as the “closing link” of the KOST stack.



TiDB is not the most widespread DBMS in the market (considering that it appeared only two years ago ), but it has potential. Giants such as Xiaomi and BookMyShow , the largest platform for booking tickets in India, are already working with it. Although it is believed that TiDB was chosen simply in order to form a beautiful abbreviation - KOST. If desired, the tool can be replaced with Cockroach, Clickhouse and other solutions.



Prospects



All solutions in the "composition" of KOST open source. Therefore, companies of all sizes and with any budgets can use the stack. KOST is expected to bring the greatest benefits to retailers and online stores. Experts say that new technologies will simplify inventory accounting through a flexible and scalable database and secure client personal data (thanks to container technology).



The LAMP stack paved the way for web applications, while MEAN, MERN, and MEVN paved the way for mobile applications. It is expected that KOST or its modifications will open up new opportunities for developers using the clouds in their projects.





Additional reading on the topic from our Telegram channel:





What we write about in the First Corporate IaaS Blog:





More articles:


All Articles