YT YouTube2Text

YouTube2Text → Transcripts → DevOps Live Interview | DevOps | Cloud | DevSecOps | SRE #interview #devops #aws #docker #cloud

DevOps Live Interview | DevOps | Cloud | DevSecOps | SRE #interview #devops #aws #docker #cloud — Transcript

by jadeja rajpal sinh · 4,957 words · 817 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:04Hi Jadeja.
0:05>> Yeah, Akash.
0:08>> So, let me open your resume once. Just
0:10give me a sec.
0:13Okay.
0:15So,
0:16till then till then please start by
0:18introducing yourself.
0:20>> Yeah. So, my name is Jadeja Rajpalsinh
0:22and I'm a DevOps engineer with 7 year
0:24plus experience in designing, building,
0:26operating scalable and highly available
0:28cloud platform.
0:29My core experience lies in implementing
0:31the key pillars of DevOps that is
0:33automation, continuous integration
0:34delivery, infrastructure as a code,
0:36monitoring, and security.
0:38Throughout my journey, I have worked on
0:40multiple cloud platform, but most
0:43extensively in the side of AWS and a bit
0:45of GCP and Azure. I have worked on
0:47designing cloud native platform,
0:49Kubernetes based environment, and
0:51end-to-end CCD ecosystem for enterprise
0:53application.
0:54And if I talk about on the side of
0:56governance, then I have implemented
0:58secure multi-cloud architecture,
1:01centralized access management solution
1:03aligned with compliance standard like
1:04GDPR and HIPAA. I also been involved
1:07into the client discussion and pre-sales
1:09activity where I need to translate the
1:11business requirement into the technical
1:13solutions.
1:14Along with the hands-on engineering, I
1:16have certification like AWS certified
1:18solution architect, Red Hat certified
1:20system administration, and Red Hat
1:22Ansible automation exam. So, I think
1:24yeah, this is just a brief about me.
1:27>> Okay.
1:28So, currently in your organization, are
1:30you working independently or as a part
1:32of DevOps team?
1:34>> Um like
1:36we are a service-based company, so right
1:38now I'm working on two projects
1:39simultaneously. In one project, I'm
1:41working as an individual contributor
1:43where I independently manage the DevOps
1:45activity. And in another project, I am
1:47leading a team of three DevOps engineer
1:49where I'm responsible for task planning
1:51and technical guidance.
1:53>> Okay. Okay.
1:55So, let's start with the questions.
1:58Suppose you need to host a new
2:00application that is expected to handle
2:03around 5 million concurrent users.
2:06What How would you design that? And what
2:09all services and tools would you use?
2:12>> Okay, so 5 million concurrent users. So,
2:14before designing the architecture, I
2:15would first understand the application
2:17requirement, whether it will be
2:19stateless, stateful. We need to also
2:22understand the expected traffic pattern,
2:24the database requirement, compliance,
2:25and latency expectation. And based on
2:28that, I would design a highly available
2:29and secure platform. So, like if I talk
2:31about the upper layer architecture, then
2:33it would be a it would have a route 53
2:36for DNS DNS, then we we would have
2:39CloudFront for CDN and caching. We could
2:41use a web and shield for security.
2:44Then to distribute the traffic, we can
2:45use load balancer. And if you're going
2:47with the containerization approach, then
2:49either we can go ahead with ECS or EKS.
2:51I would choose ECS if it like if the
2:55architecture is not that much complex,
2:57and ECS has lower operational overhead.
2:59But if you want some sort of advanced
3:01feature of Kubernetes like
3:03GitHubs, some sort of custom networking,
3:06or advanced deployment strategy, then in
3:08that case, I would use EKS one. And also
3:10at the same time, we can use Argo CD for
3:13the deployment.
3:14And for the backend, we can go ahead
3:16with Aurora and DynamoDB for the
3:19database, Redis for the caching, S3 for
3:21the static content, SQS Kafka for
3:24asynchronous processing. And there are
3:26multiple tools for the monitoring and
3:27CCD that we can set it up over there.
3:29So, it usually depends on the budget.
3:31Like if the client can't like can't pay
3:33it, then they can't have it. So, we need
3:34to understand the cost factor as well
3:36there over there. So, but we can go
3:38ahead with some sort of open source
3:40tools like Prometheus, Grafana,
3:42CloudWatch. We can go ahead with some
3:43sort of CCD tools like Jenkins, GitHub
3:46Action, GitLab CI. Yeah, so I think this
3:48would be some of my checklist that I
3:50would follow.
3:52>> Okay. Cool. Suppose you successfully
3:56build this application and your
3:58application is already in the running
3:59state, right? And you what you notice is
4:02sudden traffic spikes up to 10x, right?
4:06How would you determine whether it's the
4:08genuine traffic or a DDoS attack?
4:11>> Okay. So,
4:13as a debugging point of view, firstly I
4:15would analyze the access log to look at
4:17the source IP, request pattern,
4:19geolocation, and the request path. If
4:21the traffic is coming from diverse real
4:23users with normal behavior and valid
4:24application request, it's likely a
4:26genuine traffic spike. So, over there I
4:28would simply scale up the infrastructure
4:30using the auto scaling. But, if I see a
4:32large number of repeated requests from a
4:34suspicious IP or maybe some unusual user
4:37agent, and also at the same time if they
4:39are requesting at or targeting a single
4:41endpoint, so I would suspect a DDoS
4:43attack over there. So, to mitigate it, I
4:45would use a AWS WAF, AWS Shield, or like
4:48we can use some sort of rate limiting
4:49functions over there, and it will block
4:51the malicious IP. Like, also at the same
4:53time, we need to continuously monitor
4:55the application to ensure the legitimate
4:58users are not getting impacted over
4:59there.
5:01>> Makes sense, yeah.
5:03Um okay. So, suppose as you mentioned
5:07auto scaling in your answer, that
5:09reminds me of Kubernetes. So, have you
5:12worked on Kubernetes, yes?
5:13>> Yeah, yeah. Actually, we worked on the
5:15Kubernetes side.
5:16>> Okay. That's cool. So, suppose you have
5:19three name spaces in a Kubernetes in the
5:21same Kubernetes cluster,
5:23how would you prevent the dev name space
5:25from communicating with the prod name
5:27space?
5:28>> Okay. So, like in our case, ideally we
5:31create a different cluster for like
5:33different cluster for different
5:35environment.
5:36But, let's say if we are having all the
5:38environment into a single cluster, then
5:40over there we could use the Kubernetes
5:42network policy to restrict the cross
5:44name space communication. By default,
5:46like I would deny all the ingress and
5:48egress traffic, then explicitly we can
5:50allow only the required communication
5:52within the namespace that is required
5:54over there.
5:56>> Got it. Got it.
5:57So, have you created any, you know,
6:00deployments or services or ingress out
6:02there from the YAML file? Have you ever
6:05>> like we are creating it from the scratch
6:06even though we are working into the helm
6:08chart format, so yes.
6:10>> Okay. So, just let me know what happens
6:13internally when you run a kubectl apply
6:16hyphen f the supposed deployment.yaml.
6:18What happens internally?
6:20>> So, whenever we apply the kubectl
6:21command, so basically kubectl reads the
6:23YAML for manifest, then it sends the
6:25request to the kube API server, and then
6:28API server validates the request, some
6:30sort of authentication and
6:31authorization, and then it store the
6:33data into the etcd.
6:36Then the scheduler assign the port to a
6:38suitable worker node, then the kubelet
6:40on that specific node receives the
6:41instructions and ask the container
6:43runtime to create the container.
6:45Finally, the port start running and the
6:46container, like, sorry, the controller
6:48continuously ensure that the desired
6:50state is maintained, like, whatever the
6:51replica that you are mentioning inside
6:52the deployment file.
6:54So, usually this is how the flow works.
6:57>> Okay. So, based on your previous answer,
7:01um I assume that you're pretty much well
7:05versed with the Kubernetes architecture,
7:07no? Like, you have Have you heard about
7:10etcd on the master plane?
7:12>> Yes, yes. So, basically it's a database
7:13where it store the data of everything.
7:16>> Okay. So, like,
7:19answer this, like, why did Kubernetes
7:22choose etcd instead of any other
7:24database? They could have chosen any
7:25database, but why etcd?
7:29>> Um
7:30like, to be honest, like, I haven't
7:32looked into that depth, but my
7:34understanding is that etcd is built for
7:35distributed system, and it provides a
7:38consistency and leader election,
7:39specially the leader election.
7:41So, which are important for Kubernetes,
7:42so that is what my current
7:44understanding, but I would
7:45uh need to check into a bit more. Yeah.
7:49>> So, yeah, fine, fine, fine. Makes sense.
7:51So,
7:52>> [clears throat]
7:53>> your your you must have hosted, you
7:55know, microservice uh applications or
7:57the websites out there, right?
7:59>> Mhm.
8:00>> So, what do you think like how does the
8:03request reach your pod when a user
8:04accesses your website?
8:07>> So, like whenever user sends a request
8:08to our website, the request first
8:10reaches our reaches to the DNS. So, it
8:13can be route 53, it can be GoDaddy, it
8:15can be anything, and then it resolves
8:17the domain from there. And then it
8:18reaches like if you are using the
8:20application load balancer, then it will
8:21firstly go to the application load
8:23balancer in our case.
8:24Then it forwards the request to the
8:25Kubernetes ingress controller. And the
8:27ingress controller the ingress routes
8:29the request to the appropriate
8:31Kubernetes service, and then the service
8:33forward it it to the one of the healthy
8:35pod using the kube proxy.
8:38>> Okay. Okay.
8:40Uh let me just allow me a sec. Let me
8:42have a look at your resume.
8:45Okay. So, in your achievement section,
8:49you mentioned that you migrated 500 TBs
8:52of S3 bucket of data of S3 bucket from
8:55one AWS account to another.
8:58Mhm. Um like from the point of
9:01accomplishments, isn't that
9:02straightforward? You you could have just
9:05used AWS data sync, right?
9:08Can you just elaborate on that?
9:10>> So, even though to perform that
9:11migration activity, we used the AWS data
9:13sync. So, the migration part was never a
9:16bigger challenge for us. The real
9:18challenge was to how to very re-verify
9:21that all the data has been migrated
9:22successfully without missing any object
9:24because since the data was into the 500
9:26TB of data, so it was very difficult for
9:28us to re-verify the objects whether
9:30everything has been successfully
9:31migrated or not. So, to solve that
9:34specific issue, I I the some more
9:36services of AWS like I used the S3
9:38inventory. It basically generates a CSV
9:40format containing all the objects in a
9:42in the bucket. And I then loaded that
9:44inventory into a DynamoDB table and like
9:47then simply compared the source and the
9:49destination inventory to identify any
9:51missing object with the help of that
9:52specific DynamoDB database.
9:55>> Okay. Okay. Makes sense.
9:58Um okay. So, how like uh in your 6-7
10:02years of DevOps career, I hope that you
10:06must have built a lot of pipelines out
10:08there, right?
10:09>> Correct.
10:10>> Uh so, can you just walk me through the
10:12best CI/CD pipeline that you built?
10:16>> So, actually we I worked on multiple
10:17pipelines with different types of tools
10:19and different different type of
10:20technology, but in my recent project
10:22that currently where I'm working on, so
10:24let me walk you through over there. So,
10:26basically the approach would be almost
10:28similar for almost all the pipeline all
10:29the tools. So, usually whenever a
10:31developer pushes the code, the pipeline
10:33will automatically get triggered. It
10:35will firstly pull the latest code from
10:36the Git. It will run some unit test
10:38cases. Then it will perform some sort of
10:40code quality analysis using the
10:41SonarQube. Then while creating the
10:43image, we also use some sort of
10:45vulnerability scanning tool like Trivy
10:46and everything. Then it pushes the image
10:48to the ECR.
10:50Then for the CD part, we are having
10:52different repositories. So, over there
10:53we update the image tag in the GitHub
10:55repository.
10:57Then Argo CD automatically detect the
10:59change and it synchronize it with the
11:01our Kubernetes cluster.
11:03And then the application was uh in our
11:05in our current project, the application
11:06was deployed using the rolling update
11:08strategy. And post deployment, health
11:10checks were performed before making
11:11before marking the deployment as
11:13successful.
11:15>> Uh you mentioned uh the application was
11:17deployed using rolling update, right?
11:20>> Correct.
11:21>> Uh so, why didn't you use a canary
11:23deployment and instead you used rolling
11:25update? Because canary deployment is too
11:27much in fashion, no?
11:28>> Correct.
11:29>> So, why?
11:30Yeah.
11:30>> So, basically the thing is that we are
11:32at the stage where we are enhancing the
11:34infrastructure.
11:35So, like in the current project, like we
11:37are using the rolling update because
11:39like the application was backward
11:40compatibility, and the deployment risk
11:42was very low. So, it provided us a
11:44zero-downtime deployment with minimal
11:45operational overhead.
11:47But let's say if we have to set up a
11:48canary deployment for some sort of
11:51business-critical application or major
11:52releases, we could really
11:54implement the canary deployment. It just
11:56like we need to just
11:58do some sort of basic testing. Like the
12:00basic implementation, it will just allow
12:01us to allow give us a feature to
12:03release a new version to a small
12:05percentage of user. Then we can simply
12:07monitor the key matrices like the error
12:08rate, latency, and then we can simply
12:10gradual gradually increase the traffic.
12:12And if any issue is detected over there,
12:14then we can simply roll back into the
12:16previous version. So, if if needed, then
12:18we can also implement the canary
12:19deployment. But we are at the stage
12:21where we are enhancing the
12:22infrastructure.
12:24>> Got it. Got it. Makes sense. Yeah.
12:27So, like while creating CI/CD
12:30deployments out there, there you might
12:32have faced a lot of challenges.
12:34Sometimes some havoc, some
12:37issue in the configuration. So, let me
12:39give you a scenario. Let's say a
12:41developer accidentally commits an AWS
12:44access key
12:45and the secret key to the Git Git
12:47repository.
12:49What And that that's a blunder, no? What
12:52immediate actions would you take and how
12:55would you prevent this from happening in
12:57future?
12:58>> So, firstly, what I will do, I will just
13:00simply disable or rotate the
13:02compromised access key so that it cannot
13:04be used anymore. And then I will go to
13:07the AWS, and then we can we could check
13:09the cloud trail to just to identify if
13:11any suspicious activity has been done
13:13with the those compromised key or not.
13:16And then
13:17for the future, we can integrate some
13:19sort of Git secret or Git leak into the
13:21CI pipeline to detect whether any of the
13:23secret is getting pushed or not before
13:25the code is getting merged.
13:26And then like at the like the best
13:28practices is we should store some sort
13:30of sensitive credential into the AWS
13:33secret manager or we could save it to
13:34HashiCorp Vault as well.
13:36And also, at the same time, we need to
13:38educate the developers on secure
13:40credential management as a part of the
13:42development workflow.
13:44>> Okay. Okay, makes sense.
13:47So, sup- let's suppose
13:51like by using your CI/CD pipeline or
13:54something, uh you deployed a new version
13:56of the application, right? The
13:58deployment is showing successful, that
14:00means the green deployment, right? So,
14:02it's green on your side, but the
14:04customers are, you know, getting a lot
14:06of 502 bad gateway errors.
14:09Right? So, how would you troubleshoot
14:10that?
14:12>> So, firstly,
14:14uh okay, the deployment goes green, but
14:15still the users are impacting or getting
14:17502 errors.
14:18>> Yeah, 502. Yeah, yeah.
14:19>> So, maybe firstly, I will check the logs
14:21of the pod, and maybe I will check
14:23whether that pods are up and running or
14:25not. Maybe I would also check some sort
14:26of readiness and liveness probe as well.
14:29And then, I will verify whether the
14:30service is correctly pointing or
14:31forwarding the traffic to the healthy
14:33pods or not.
14:34Then, even though if that is also
14:35correct, then I will check the ingress
14:37controller or the application at the LB
14:39logs to identify whether the from where
14:41the 502 is being generated. Then, also I
14:44will at the same time, I will I will try
14:45to check some sort of application logs
14:47for any sort of startup or runtime
14:49error.
14:50Then, like, I will also ensure that the
14:52the application is listening on the
14:53correct port and the service target port
14:56matches the container port that we
14:58defined inside the deployment file.
15:00And even though if the issue is still
15:02persistent over there, so my first
15:04preference would be to roll back it to
15:05the
15:06previous stable version to restore the
15:08services while investigating the root
15:10cause. And we can we could do it onto
15:12the lower environment. So, firstly, I
15:13will roll roll back it to the previous
15:15version, so that at least the end
15:16customer would not impact anymore.
15:19>> Got it. So, like, suppose as you
15:22answered, you will check this you will
15:24check a lot of things. So, like during
15:27your investigation of the checks, you
15:30find that the suppose you during your
15:32investigation of the checks, you find
15:33that the 502 error is caused by the
15:36database, right? And the database CPU is
15:39consistently above 90% since you made
15:44the deployment. And one there's one
15:46query that is being executed hundreds of
15:48times per minute because of the because
15:52of the particular application
15:53functionality that's assume. So, how
15:55would you solve this?
15:57>> So, the firstly I would do is to confirm
16:00whether the repeated query is actually
16:01necessary or it's an application issue.
16:04Let's say if the data doesn't change
16:05frequently like whenever user is
16:07requesting that specific query. So, I
16:09would introduce Redis as a caching
16:11layer. So, instead of hitting the
16:12database directly for every request, the
16:14application firstly it would go to the
16:16Redis and if the data is available it
16:17would return the response directly from
16:20there. If the data is not present in
16:21Redis, it would query the database and
16:23then store the result in Redis for the
16:25subsequent request. So, like this would
16:27significantly reduce the number of
16:30database queries. It will also reduce
16:32the CPU utilization. So, maybe this is
16:35how we could improve the response time
16:37for the request.
16:39>> Okay.
16:41Okay.
16:42So, have you worked on Terraform?
16:44>> Yes, I'm working on Terraform. Yeah.
16:46>> Okay. So, how would you structure
16:49Terraform code for large enterprise
16:50project with multiple environments like
16:53dev, prod, and stage?
16:56>> So, we can create a modular structure in
16:58which like we can create separate module
17:00folder for component like VPC, EKS, RDS,
17:03or whatever the services that we are
17:05using into our current project. And then
17:07there would be a separate environment
17:08folder such as dev, stage, prod in which
17:10where each environment will call the
17:12same module with the different variable
17:14values.
17:15So, it would like also at the same time
17:16each environment would have its own back
17:18end and state file to keep the interest
17:20isolated. And then we could simply
17:22integrated that Terraform with a CCD
17:24pipeline to so that every infrastructure
17:26changes goes through review, validation,
17:28and approval being before being getting
17:30applied over there.
17:31So this is what we can do here.
17:33>> Okay. So while working with Terraform,
17:36Terraform being too sensitive and you
17:39know, you you need to do your work in a
17:41good fashion, right? So
17:45you must have faced some kind of you
17:47know, issue while issue as in like
17:50suppose I'll give you an example. Like
17:52what like what would happen if someone
17:54manually changes the infrastructure in
17:56AWS outside of Terraform? What would
17:59what like what happens in that case?
18:01>> Let's say
18:02if they made some sort of changes into
18:03the AWS cloud and if we are not having
18:05that code into our Terraform, then
18:07whenever we will try to run the
18:08Terraform plan, the Terraform will
18:10detect the drift and shows the
18:11difference between the desired state and
18:13the current state. So like it depend on
18:15the change. Like if the manual change is
18:16not part of the Terraform code,
18:17Terraform apply may revert it back to
18:19the desired state. But let's say if the
18:21manual change is needed into our current
18:23code right now. So I would firstly
18:25review it and then update the Terraform
18:26code so that the infrastructure and the
18:28code remains in sync. So this is what
18:30this is how we can solve it.
18:33>> Okay.
18:34So
18:35when do you use depends on? Can you just
18:37give me an example?
18:39>> So like
18:40I use depends on when I want to make
18:43one resource is created only after
18:46another resource is ready. So there is a
18:47dependency. Over there we could get So
18:49if I give you a small example, then
18:51let's say we create a private route
18:53while creating VPC, right? So I want to
18:56ensure that the net gateway because
18:58inside the route we attach the net
18:59gateway to the routes. So I want to
19:01ensure that the net gateway is fully
19:02created before the route points to it.
19:04So in such case I can accidentally use
19:06the depends on to guarantee that the
19:08creation order.
19:10>> Okay. Okay.
19:12So let's come back to AWS maybe just a
19:15few questions on AWS.
19:17What would happen if an entire AWS
19:21region goes down and how would you
19:23recover your application?
19:25>> Like we are already deploying our
19:27application with the help of
19:28multi-region. But even though in our
19:31current project since we are working
19:32with US based client, so we perform some
19:35sort of disaster recovery drills every 3
19:37months to ensure our recovery process
19:39actually works. So, there are multiple
19:41steps that we need to follow. So, we are
19:42keeping a backup of our database like we
19:45use the database snapshot. It will
19:47automatically copy like it will
19:49automatically create the snapshot and
19:50there is lambda function which will move
19:51the snapshot from one region to another
19:53one another one, sorry. And then we also
19:56enable the S3 cross region application
19:57that is simply CRR so that our
19:59application data is already available in
20:01the DR region. And since our
20:03infrastructure is managed through
20:04Terraform, so we can quickly recreate
20:06the infrastructure in the secondary
20:07region. That won't be an issue much
20:08more.
20:09And once the application is up and
20:11ready, like we can simply update the
20:12route 53 to route the traffic to the
20:14healthy region.
20:15And we have already tested this DR
20:17process multiple times in our current
20:19project. So, like right now like we are
20:22much confident that we could cover the
20:24application like we can recover the
20:25application within the agreed RTO or the
20:28RPO instead of figuring it out at the
20:30run time.
20:31>> Okay.
20:33Okay, okay.
20:34I was just going through your resume. I
20:37find one thing very interesting.
20:40You worked on
20:43>> [clears throat]
20:43>> HIPAA and GDPR compliances, right?
20:46>> Correct.
20:47>> So, what was your exact role in this?
20:50Can you just mention?
20:51>> So, basically I was working for a client
20:53and they are into the health care
20:54sector. So, since they are applying
20:56around the data of the patient. So, they
21:00need to be a bit more
21:02concerned about the security part. So,
21:04like currently they They into the Europe
21:06under the GDPR compliances. So, when
21:07they expanded their business to the US
21:09so they also needed to comply with the
21:11HIPAA requirement. So, as a part of the
21:13initiative, our our focus was on
21:15strengthening the infrastructure from a
21:17security and compliance perspective. So,
21:20if I give you a small example, there
21:21were a lot a lot of checklist over
21:23there, but if I give you a small
21:24example, then uh we ensured the private
21:26S3 bucket were accessed through VPC
21:28endpoint instead of the public internet.
21:30Then we are implementing some sort of
21:31governance policy in our community
21:32cluster to uh to enforce the secure
21:35deployment uh standards. Then we make
21:37sure that the PI and the PHI data was
21:39never written to the application logs by
21:42masking or removing the sensitive
21:43information from that logs.
21:45Then we are also like we also enforce
21:47the encryption. We also use some sort of
21:49least privileges I am access and
21:51regularly we review the security and
21:53everything, yeah.
21:55>> Okay. Okay.
21:57So, let [clears throat] me give you a
21:59real-life scenario, right?
22:01>> Mhm.
22:01>> So, uh right now FIFA is uh too much
22:05trending and the ads and the banners in
22:07the FIFA. You must have heard that the
22:09in different countries they show
22:11different kind of ads or different kind
22:12of pages, right? So,
22:15>> [clears throat]
22:15>> like I would like to question on that in
22:17CloudFront, if you want to show
22:19different pages based on the user's
22:21country, how would you implement that?
22:25>> Um different pages based on the their
22:28country, right?
22:29>> Yeah.
22:30In CloudFront.
22:31>> Oh, CloudFront, right.
22:32So, there would be multiple ways of
22:34doing it. Uh like the firstly, the
22:36simplest would be like we can write up a
22:37small logic inside the application that
22:40will handle the uh
22:42logic which will uh move the traffic
22:44from one page to different one according
22:46to the user's country. But let's if you
22:48are specifically going with the
22:50CloudFront, so there is a there is one
22:51more option that we can do it uh
22:53CloudFront
22:55provides us the CloudFront function.
22:57So, we could use the CloudFront
22:58functions over there and there is a uh
23:00header like we can use uh, viewer
23:02country header, like I don't exactly
23:04remember the header, but there is some
23:05sort of CloudFront viewer country, some
23:07sort of header like that.
23:07>> Yeah.
23:08>> And then we can write up a small
23:09CloudFront function, and that it will
23:10automatically inspect that specific
23:12header, and rewrite the request URI or
23:14the route the users to a different uh,
23:17content over there. Let's say if the
23:18user is coming from India, so it will
23:20uh,
23:21access on the server page like with some
23:23sort of landing page. So we could create
23:25a CloudFront function. Then there are
23:27multiple ways of doing it. So even
23:28though we could implement some sort of
23:29Lambda function through which
23:30Lambda@Edge which will handle this
23:32specific logic. So there are multiple
23:34ways of doing it.
23:36>> Mhm. Okay.
23:38Okay.
23:40So
23:41uh, have you worked in multi-cloud
23:42environments?
23:44>> Yeah, so like uh, like within this uh, 7
23:47or 6 years of my cloud journey, so I got
23:50a chance to work on different different
23:51projects. I worked on AWS, GCP, Azure. I
23:54did a migration of like multiple times I
23:56did a migration of from GCP to AWS as
24:00well. I did a lot of uh, I have created
24:03a lot of pipeline in terms of Azure
24:04DevOps as well. So like it's it's like
24:06they have worked on almost all the cloud
24:08provider, but most extensively on the
24:10side of AWS.
24:12>> Most ex- And I also see that you have
24:16uh,
24:17AWS certification, right?
24:18>> Correct. Correct. So yeah, it's a
24:19professional one.
24:21>> Yeah, it's a professional one. So when
24:23did you get this?
24:25>> So I did that in the year somewhere
24:27around 2023. So maybe it would be in
24:30some sort of expiration state or maybe
24:32like it would be a bit closer to the
24:34expiration.
24:35>> Yeah, it's like fine. It's like
24:38Okay.
24:39Uh, let me go through your resume once
24:42more.
24:43Just allow me a second.
24:45Okay. So I was Yeah, I was asking that
24:49Mhm. Like, [clears throat] can you just
24:51tell me more about Kibana, like a little
24:53about Kibana?
24:55>> So right now we have used the Kibana in
24:56our current project, so it is helping us
24:58to uh,
25:00strengthening the security of the cabin.
25:01So, let me give you a small example. So,
25:03we have like it's just a YML file in
25:04which you could create any sort of
25:06policy that you wanted to use in your
25:07Kubernetes cluster. In our case, like
25:09I'm using like this is one of the policy
25:11that we are using.
25:12So, that we have mentioned the account
25:14ID of the AWS account account AWS
25:15account ID. So, that whatever the image
25:18that is present into that specific AWS
25:20account ID, only those deployment would
25:22be able to deploy. They won't they won't
25:24be able to deploy any sort of basic like
25:26Ubuntu, Nginx, Python. They won't be
25:28able to deploy anything. Only the images
25:30that is present on that specific AWS
25:32account ID. So, this is how we are
25:34restricting to use
25:36secure images that is present on our AWS
25:38account. Yeah.
25:40>> Okay.
25:42Okay, makes sense. Um
25:45yes, I guess I covered most of the
25:47points.
25:49Um yes, so yep, I'm good from my side.
25:53Um maybe if you want to have some
25:54questions, please spam them.
25:56>> So, actually I don't have any specific
25:58question, but just wanted to understand
25:59that what kind of project or business
26:01domain that your organization is working
26:02on and also maybe some bit idea about
26:04the tools and the technology.
26:07>> Okay. So, in our organization, we work
26:11with multiple clients across multiple
26:12domains such as, you know, health care,
26:15fintech, a lot of e-commerce clients out
26:17there, retail clients out there. So, the
26:20technology stack varies like depending
26:23on the client based on the client
26:24requirement. But most of our project
26:26revolve around AWS, Kubernetes,
26:28Terraform, Docker, GCP, and monitoring
26:30tools.
26:31So, as a DevOps engineer, you'll likely
26:34work on different projects and, you
26:37know, get exposure to multiple
26:38technologies over the same time. Yeah.
26:41I guess I answered your question.
26:43>> Yeah, yeah. Yep, perfect. I think so,
26:45yeah, that is the only question that I
26:47have right now.
26:48>> Okay. Okay.
26:50So, nice meeting you, Jadeja. I'll just
26:53pass on my feedback to the HR. She'll
26:55get back to you.
26:55>> Sure. Thank you so much. Thank you so
26:57much. Yeah, have a nice day.
26:58>> Thank you.

About this transcript

This page contains the full transcript of DevOps Live Interview | DevOps | Cloud | DevSecOps | SRE #interview #devops #aws #docker #cloud by jadeja rajpal sinh, generated from the public captions YouTube serves with the video. The transcript has 4,957 words across 817 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to translate the transcript into 100+ languages.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.