1 of 53

DEVOPS & SRE THROUGH THE AGES AND WITH AI HELP

EPAM Proprietary & Confidential.

EPAM Proprietary & Confidential.

1

2 of 53

GLORY AND MEMORY TO OUR HEROES

EPAM Proprietary & Confidential.

2

3 of 53

Agenda

Introduction

Old School

  • Bio
  • Timeline of pitch
  • Experience matters
  • Approaches
  • Experience
  • Technology and Metrics

Mid School

  • Cloud & Approach
  • Automation & Automatisation
  • Measurement

Nowadays

  • Speed vs Value
  • How to measure or what to
  • AI and its value
  • More & More interesting with use cases

01

02

04

03

EPAM Proprietary & Confidential.

3

4 of 53

INTRODUCTION

01

EPAM Proprietary & Confidential.

4

5 of 53

PROFILE SUMMARY

  • 18-20+ years of production experience in IT
  • 16+ years of Systems engineering
  • 13+ years EPAM experience
  • Managerial experience (leading teams starting from 2012, RM)
  • Core functions/roles Architecture, Technology, Engineering, Service and Delivery within Consultancy
  • Hybrid skills across Systems Engineering, Digital and Cloud Solutions and not limited to
  • The goal is to Automate everything and Monitor everything (all the rest AI will do for us or not)
  • DevTestSecOps best practices as well as strategic view
  • Head of Cloud Competency Center at Epam

KOSTIANTYN SEVERENCHUK

Director, Technology Solutions

Lead Systems Architect

Ukraine, Lviv

EPAM Proprietary & Confidential.

5

6 of 53

OLD SCHOOL

02

EPAM Proprietary & Confidential.

6

7 of 53

EPAM Proprietary & Confidential.

EPAM Proprietary & Confidential.

7

8 of 53

EPAM Proprietary & Confidential.

8

9 of 53

WHAT WE HAVE HAD 10 YEARS BACK

OLD SCHOOL

Infrastructure as a Service (IaaS) by major players like AWS, Microsoft Azure, and GCP.

Software as a Service (SaaS) as baseline of distributed model

Limited Hybrid Solutions.

Focus on Virtualization

Basic Security Measures as well as concerns.

Cost Efficiency.

Development of APIs.

Limited Data Analytics.

Regulatory Challenges.

EPAM Proprietary & Confidential.

9

10 of 53

STAYTMENT

OLD SCHOOL

Overall, cloud technology ten years ago was characterized by rapid innovation and growing acceptance, but it was still maturing in terms of security, integration, and comprehensive service offerings.

EPAM Proprietary & Confidential.

10

11 of 53

INTRODUCTION

EPAM Proprietary & Confidential.

11

12 of 53

STAYTMENT

OLD SCHOOL

Overall, the landscape of infrastructure automation ten years ago was characterized by a mix of traditional tools and emerging technologies. Organizations were beginning to recognize the importance of automation in improving operational efficiency, reducing human error, and enabling faster deployment cycles. As cloud computing continued to evolve, the integration of these automation tools became essential for managing increasingly complex IT environments

EPAM Proprietary & Confidential.

12

13 of 53

OLD SCHOOL

Tool

Estimated Time

Docker

15-30 minutes

Kubernetes

1-2 hours

Jenkins

30-60 minutes

Travis CI

15-30 minutes

Nagios

1-2 hours

ELK Stack

2-4 hours

VMware

1-3 hours

OpenStack

4-8 hours

SaltStack

1-2 hours

Cisco ACI

2-4 hours

Ansible

30-60 minutes

Veeam

30-60 minutes

SUMMARY OF ESTIMATED TIMES

EPAM Proprietary & Confidential.

13

14 of 53

STAYTMENT

OLD SCHOOL

The estimated times for infrastructure provisioning 10 years ago highlight the varying complexities and efficiencies of different tools. While some tools like Docker and Travis CI offered quicker setups, others like OpenStack and ELK Stack required significantly more time due to their complexity. This landscape has evolved, with many tools becoming more user-friendly and efficient over the years.

EPAM Proprietary & Confidential.

14

15 of 53

MID SCHOOL

03

EPAM Proprietary & Confidential.

15

16 of 53

WHAT WE HAVE HAD 5 YEARS BACK

MID SCHOOL

1.Multi-Cloud and Hybrid Cloud

2. Containers and Kubernetes

3. Serverless Computing

4. Edge Computing

5. AI and Machine Learning in the Cloud

6. Advanced Security and Compliance

7. Cloud-Native Applications

8. Managed Services

9. Global Expansion of Data Centers

10. Cost Management and Optimization

EPAM Proprietary & Confidential.

16

17 of 53

INTRODUCTION

EPAM Proprietary & Confidential.

17

18 of 53

STAYTMENT

MID SCHOOL

Overall, cloud technology five years ago was characterized by increased complexity, flexibility, and maturity, with a focus on enabling scalable, efficient, and innovative solutions.

EPAM Proprietary & Confidential.

18

19 of 53

EPAM Proprietary & Confidential.

19

20 of 53

SUMMARY OF ESTIMATED TIMES

MID SCHOOL

Tool

Estimated Time

Docker

15-30 minutes

Kubernetes

1-2 hours

Jenkins

30-60 minutes

Travis CI

15-30 minutes

Nagios

1-2 hours

ELK Stack

2-4 hours

VMware

1-3 hours

OpenStack

4-8 hours

SaltStack

1-2 hours

Cisco ACI

2-4 hours

Ansible

30-60 minutes

Veeam

30-60 minutes

REDUCED 50% COMPARED TO 10 YARS AGO

EPAM Proprietary & Confidential.

20

21 of 53

STAYTMENT

MID SCHOOL

Five years ago, infrastructure automation was characterized by a diverse set of tools and technologies aimed at improving efficiency, consistency, and reliability in IT operations. The rise of cloud computing and containerization significantly influenced the landscape, leading to the adoption of Infrastructure as Code (IaC) practices and the integration of automation into various aspects of IT management. Organizations were increasingly looking to streamline their processes, reduce manual intervention, and enhance their ability to scale operations effectively.

EPAM Proprietary & Confidential.

21

22 of 53

NOWADAYS

04

EPAM Proprietary & Confidential.

22

23 of 53

EPAM Proprietary & Confidential.

23

24 of 53

WHERE WE ARE RIGHT NOW

NOWADAYS

1. Infrastructure as a Service (IaaS)

2. Platform as a Service (PaaS)

3. Software as a Service (SaaS)

4. AI & Machine Learning Integration

5. Data Storage and Management

6. DevOps & Cloud Native

7. Security & Compliance

8. Internet of Things (IoT)

9. Cost Management & Monitoring

10. Managed and Fully Managed Services

11. Global Research and Scalability

EPAM Proprietary & Confidential.

24

25 of 53

STAYTMENT

NOWADAYS

Modern cloud technology is characterized by its flexibility, scalability, and the integration of advanced technologies like AI and edge computing. It empowers organizations to innovate rapidly, scale on-demand, and optimize costs, making it a foundational component of the digital transformation journey.

EPAM Proprietary & Confidential.

25

26 of 53

EPAM Proprietary & Confidential.

26

27 of 53

EPAM Proprietary & Confidential.

27

28 of 53

NOWADAYS …

BUT WITH AI �EXAMPLES!

Title Goes Here

EPAM Proprietary & Confidential.

28

EPAM Proprietary & Confidential.

28

29 of 53

THE CHALLENGE - DEVELOPERS' DILEMMA

NOWADAYS

Developers dedicate significant time to routine tasks, such as writing unit tests, generating documentation, and maintaining build and deploy pipelines. These tasks, while seemingly mundane, are crucial for ensuring the code they produce adheres to high-quality standards, involving meticulous vetting for syntactic and semantic accuracy, bugs, and other potential vulnerabilities.

Fulfillment of Routine Task

These routine tasks, while undeniably critical to the software development life cycle (SDLC), can often divert developers' attention from the more strategic aspects of their work, such as business logic and innovation. This diversion can stifle creativity and slow down productivity, leading to longer project turnaround times and increased development costs.

Impact on the Overall�Productivity

To exacerbate the issue, quality assurance often becomes an afterthought due to this hefty workload. Developers are pressed for time and frequently rush through testing, increasing the risk of errors, system vulnerabilities, and other problems that could have severe consequences for the end product, such as compromised security, poor user experience, and even legal issues.

Quality Assurance Neglect

EPAM Proprietary & Confidential.

29

30 of 53

EPAM Proprietary & Confidential.

30

31 of 53

USE-CASES FOR AI

We had 40 LLMs, 3 public cloud platforms, 11 microservices and a variety of API…

Can’t say all of if was necessary, but it’s hard to stop getting into AI and ML.

EPAM Proprietary & Confidential.

31

32 of 53

CASE 1

4.1

EPAM Proprietary & Confidential.

32

33 of 53

JDM-BOT – AI-POWERED

CASE 1

Help developers to increasing their productivity and focus on business logic instead of code quality.�

How it works?

Bot implements necessary additions to pull request, automating tests creation and most of YAML / JSON based metadata generation, turning a piece of code into well-shaped building block for software solution.

Key use-cases includes generation of

    • UNIT tests and test cases
    • Low-level documentation
    • Code build and deploy pipelines
    • Metadata and interface specifications

EPAM Proprietary & Confidential.

33

34 of 53

How it works

CASE 1

EPAM Proprietary & Confidential.

34

35 of 53

Solution concept

CASE 1

EPAM Proprietary & Confidential.

35

36 of 53

Technical stack

CASE 1

What did we try to use (Nov 2023)

What are we using, as for now (Feb 2024)

  • GCP Cloud Functions to run application code
    • Azure / AWS portable
  • GCP Cloud Build for CI /CD
  • GitHub (and Azure DevOps a bit) for code storage

plus

EPAM Proprietary & Confidential.

36

37 of 53

Expectations and reality

CASE 1

Expectations

Reality (December 2023)

EPAM Proprietary & Confidential.

37

38 of 53

CASE 2

4.2

EPAM Proprietary & Confidential.

38

39 of 53

AI Powered CI/CD

CASE 2

InfraPhoenix is the product that brings a unified CI/CD approach with testing best practices out of the box and helps to manage infrastructure with a focus on code quality, security, and is AI-powered for supporting processes.

For Terraform code

  • Errors processing via PR with fixes
  • Regula warnings processing via PR comments

For applications code

  • Unit tests generation via PR comments
  • Source code explantation via PR comment
  • Code comments composition via PR comments
  • SonarQube quality gate issues handling via PR with fixes

EPAM Proprietary & Confidential.

39

40 of 53

AI INTEGRATION CASES V2

4.3

EPAM Proprietary & Confidential.

40

41 of 53

STATUS IMPLEMENTED

AI INTEGRATION CASES

For Terraform code

  • Unit tests generation via PR comments
  • Source code explantation via PR comments
  • Code comments composition via PR comments

For applications code

  • SonarQube quality gate issues handling via

PR with fixes

SonarQube

  • Errors processing via PR with fixes
  • Regula warnings processing via PR comments

1

2

3

AI

EPAM Proprietary & Confidential.

41

42 of 53

AI INTEGRATION CASES

SOURCE CODE EXPLANATION VIA PR COMMENTS 

STATUS IMPLEMENTED

EPAM Proprietary & Confidential.

42

43 of 53

STATUS IMPLEMENTED

ERRORS PROCESSING VIA PR WITH FIXES

FOR TERRAGRUNT WORKFLOW 

AI INTEGRATION CASES

EPAM Proprietary & Confidential.

43

44 of 53

REGULA WARNINGS PROCESSING FOR ACCELERATOR IAC 

Tools

  • Regula – implemented due to poor documentation and absence of examples in output
  • TFLint/TFSec - not applicable, by default provides a real code sample with resolution

STATUS IMPLEMENTED

AI INTEGRATION CASES

EPAM Proprietary & Confidential.

44

45 of 53

AI INTEGRATION CASES

UNIT TESTS GENERATION VIA PR COMMENTS 

STATUS IMPLEMENTED

EPAM Proprietary & Confidential.

45

46 of 53

AI INTEGRATION CASES

CODE COMMENTS COMPOSITION VIA PR COMMENTS 

STATUS IMPLEMENTED

EPAM Proprietary & Confidential.

46

47 of 53

AI INTEGRATION CASES

SONARQUBE QUALITY GATE ISSUES HANDLING FOR APPLICATIONS VIA PR WITH FIXES 

SonarQube quality gate issues handling for Accelerator IaC via PR with fixes – in progress

STATUS IMPLEMENTED

EPAM Proprietary & Confidential.

47

48 of 53

THE PRODUCT HAS BEEN DEVELOPED ON PUBLIC EPAM GITHUB REPO SINCE JULY 2023 UNDER APACHE 2.0 OPEN-SOURCE LICENSE.

EPAM Proprietary & Confidential.

48

49 of 53

SDLC stage

Saved Time (approximately man-hors by %)

Requirement Analysis

30-50%

Design

20-40%

Implementation(Codding)

30-50 %

Testing

40-60 %

Deployment

50-70 %

Maintenance

30-50 %

Documentation

40-60 %

AI has notably reduced the time required for various phases of the Software Development Life Cycle (SDLC) compared to traditional, non-AI-assisted methods.

EPAM Proprietary & Confidential.

49

50 of 53

EPAM Proprietary & Confidential.

50

51 of 53

REGISTRATION FOR RAFFLE

EPAM Proprietary & Confidential.

51

52 of 53

Thank you!

For more information, contact

Kostiantyn Severenchuk

Director, Technology Solutions

This presentation has been created with help of:

  • AWS Q
  • Chat GPT 4 and sometimes with GPT 3.X
  • Copilot
  • Polished by designer & Fixed by me

EPAM Proprietary & Confidential.

52

53 of 53

EPAM Proprietary & Confidential.

53