DevOps/SRE Interview Question #41 System Design for a Service That Spikes

DevOps/SRE Interview Question #41 System Design for a Service That Spikes
Transcript:
0:01 Hello and thank you for clicking on my video. If you like the content I'm producing, please subscribe. That really helps me out a lot.
0:08 Let's get started. Today we're going to do one more interview questions for DevOps slash SRE position. This one's, like, a little bit more towards junior positions because the question is pretty common and somewhat introductory about how anything with DevOps would work.
0:29 I've been getting a lot of requests for interview questions for junior-level positions so I will be doing more of those.
0:37 So today's question is, your company has a website that needs to be reliable and highly available during a holiday season where the traffic will spike a lot.
0:50 It's currently using a microservices based architecture using Docker on Kubernetes and it has web servers, databases and various other microservices as middleware.
1:03 How do you design this so that it will withstand the spike of holiday traffic. So if you want to think about it, pause the video and I will give you one answer in just a second.
1:20 Alright, there are many ways you can answer this question, of course. But the interviewer is looking for some commonalities, some critical component of the infrastructure design in this matter.
1:33 So some of them are road balancing, so that when the traffic speaks up, it doesn't just go to one container.
1:40 It goes to multiple pods. Second, auto scaling. As traffic goes up, your services should scale up. Multiple pods, more pods should be scaled up almost instantaneously based on some criteria.
1:57 This will make all the services, for example, frontend web services or middleware services, scale up and down as more requests keep coming in.
2:10 The question mentioned that you also have databases. So you have to use database that can also scale and be highly available if one, one container goes down or virtual machine goes down.
2:23 So MySQL and Postgres databases are very common these days. So any one of those would work on top. A couple of these design things, you should also mention a couple of things that a experienced DevOps engineer would answer.
2:40 You need to have service monitoring. Most common these days are Prometheus for matrix gathering and Grafana for dashboard. So that if there is a problem, you will know and you can fix it right away.
2:55 You also need even though it doesn't really ask for it in the question, you should also mention a CICD pipeline so that before the Before.
3:04 Of Spike or during the Spike, if you need to, you can do more frequent and reliable releases through this pipeline and that also gives you a chance to roll back if things don't work out.
3:21 I hope you found this video useful and I will be doing more of these interview caution type videos in the future and I will see you in the next one.