21 Technologies to Explore & Scale in 2021

Earlier this year we launched TechCompass to research, curate and apply the technology trends across several technology domains like User Experience, AI, Cloud, Data, API’s, Cybersecurity and Low Code No Code platforms. As we come towards the end of this year, I have picked 21 existing and emerging technologies that are either becoming dominant or have the potential to become dominant in the next 1-2 years and are worth learning, exploring, experimenting and scaling in 2021.

1.     Micro UI’s: Microservices are today the norm across most of the enterprises and its now time to start unbundling the user experience layer using the Micro front end design pattern. Unbundling UI will help teams to develop Micro UI’s independently that can then be composed and reused to build UI faster. The latest release of Webpack5 and its module federation feature makes it relatively easier to build micro front ends.

2.     PKC and Password less Authentication: With the number of unique digital touchpoints continuously increasing, remembering and managing passwords is a hassle. Fido2 set of specifications and W3C Web Authentication will make it easier to authenticate in the mobile and desktop environments. Good idea to explore and scale it across digital touchpoints

3.     HTTP/3 QUIC: The HTTP Version 3 currently in the internet draft stage will bring about a significant change by moving away from the TCP protocol to UDP based QUIC transport protocol that will enable stream multiplexing and low latency connections. Popular browsers have started supporting it and it is a good time to start planning for it

4.     Go and Rust: We rewrote one of our existing platforms using Go language and were able to reduce the memory footprint significantly by 10x and were able to improve startup performance by 2x and reduced build time by 3x. In the recent developer survey by stackoverflow, Rust was recognized as the most loved language and is memory efficient. It is worth exploring these languages for memory efficiency and high throughput use cases.

5.     MLOps: DevOps has become the norm across most of the enterprises and are now starting to embed security and privacy policies into their pipelines. However, as you scale the AI investments, in order to accelerate and scale Model lifecycle management, invest in implementing and maturing the ML operations (MLOps).

6.     NLQ: It’s time to start moving away from the canned dashboards and reports to a more intuitive way of asking questions (like google search) that matter to the business and for making decisions. At Infosys we have been experimenting and working on enabling it through knowledge graphs, embedded analytics and open-source Rasa with good results.

7.     Graph Database: In almost all the enterprises, data is not always connected in a meaningful way to enable searchability and visibility. Like digital natives’ explore and build graphs to create a network of consumers, customers, products, services and operations using graph databases like Neo4J, ONgDB, AWS Neptune etc.

8.     Browser & Cloud based IDE:  Browser based environments have been traditionally used for learning and coding assessment. However, with the emergence of new age browser-based IDE’s like beta release of Visual studio codespaces and cloud-based IDE’s like AWS Cloud9, Code Tasty explore using them to jumpstart development and projects.

9.     Neural Machine Translation: NMT is a technique that has been used for language translation and at Infosys as part of a database migration exercise, we recently explored using it for translating the SQL queries written for one of the proprietary databases to an open-source database. Using translation pairs from a dataset of 27,000 SQL queries, the Query translator was able to achieve 91% accuracy in query conversion. Facebook also published a transcoder and paper for migrating from legacy languages like Cobol. Explore this space for migration and modernization projects.

10.  Experience Design as Code: Translating visual designs and specs into code is an effort intensive activity and we are now starting to see the emergence of AI powered tools like the MSFT sketc2code, sketch2react translate UX designs done on sketch into html and react code. This space is still evolving but worth keeping an eye on.

11.  Observability @ Scale: The state of observability is summed up very well in the Catchpoint SRE 2020 report, ‘observability components exist, but observability does not’. As part of live enterprise transformation, we have seen that comprehensive observability across the layers covering usage, consumption patterns, the entire technology stack and the infrastructure helps become more resilient and to continuously learn and improve. It’s time to drive observability @ scale.

12.  Kubernetes Ecosystem: Kubernetes is becoming the norm for the cloud first architecture and implementations across most enterprises and becomes even more critical in a hybrid, multi cloud setup. Some of the projects in the Kubernetes ecosystem that we have found interesting and are implementing at scale are Kubeflow for developing and deploying ML services and KubeEdge for running containerized apps on the edge.

13.  Differential Privacy: With more and more enterprises now instrumenting, capturing and consuming data for consumer, customer and employee analytics and using these datasets for training and developing AI services, embedding and using differential privacy and investing in it has become essential. Differential privacy is already used by Apple and Google, and Google and IBM have open sourced it as well.

14.  TICK Stack:  TICK stack is a collection of open-source software’s that has been around for a while, but we have found it to be very useful in implementing large scale monitoring, eventing and metrics. It’s interesting to explore and scale as it uses a time series database, enables SNMP based monitoring, timeline-based exploration and graphs.

15.  Data Egress: As more and more enterprises embrace multi cloud and hybrid cloud ecosystem moving terabytes and petabytes of data into the cloud, it’s important to understand the cost implications of moving data into cloud (knows as ingress and generally free) and moving your data out of cloud (known as egress) that is typically charged. Plan for Data egress upfront and include the costs in your business case.

16.  Low Code No Code (LCNC) AI Platforms: LCNC platforms have been used by citizen developers for the last few years and now we are starting to see the emergence of LCNC platforms like Google Clouds AutoML, Uber’s Ludwig, Baidu’s EZDL enabling basic AI services for vision, speech and recommendations. This is still a fast-evolving area and good to keep an eye on.

17.  Tiny AI: Tiny AI is an effort led by Tiny ML and technology majors to reduce the overall size of algorithms and improve performance so that more and more AI services can be run independently on the devices (think 3.5B+ smartphones globally) without having to send significant data to the cloud. This coupled with speciality AI chips designed to consume less power and deliver more computational power is likely to help scale AI usage and use cases significantly.

18.  WebRTC: This year when all of us had to move to an anytime anywhere operating model, all the collaboration had to be done through digital platforms. WebRTC an open-source project that provides browsers and mobile apps real-time audio, video, chat and screen sharing capabilities in a browser first manner has been a saver and we developed interesting collaboration platforms and features using WebRTC. Explore and scale usage of WebRTC.

19.  Cloud Native Network Functions (CNF): Containers have started moving into the networking areas with CNF’s providing an ability to deploy and run network functions within containers. CNF’s promise to be more efficient and deliver higher development velocity and better performance than VNF’s though they are likely to co-exist with VNF in the near future. Interesting space to watch for developments from CNCF and LFN.

20.  GraalVM:  It’s a polyglot virtual machine that supports JVM based languages like Java, Scala, Clojure as well as supports Java script, Python and Ruby. It enables calls to be made to programs written in any of the supported languages without the foreign language overhead. It is memory efficient, performant and for those having java application landscape its worth exploring and adopting 

21.  Fluid Framework: Microsoft recently open sourced the Fluid Framework  that provides a distributed data structure enabling multiples users to collaborate and perform activities like editing content in real-time, that is then immediately made visible to everyone. While it is still early days, but the framework provides powerful live collaboration capabilities that can make the user experience lot better. Keep an eye on this and explore…

Hope you find these interesting to explore further. If you have any additional recommendations or suggestions on technologies that are ready for production use, please share.