Show HN: I made DevOps as productized service | Hacker News Reader

Show HN: I made DevOps as productized service

stunningdevops.com

5 pointsrobustpython16 comments

I'm normal contractor programmer, for the past few years doing mainly DevOps on AWS. I noticed lot of smaller companies don't have enough work for full-time devops person nor they have cloud/aws knowledge on the team. Lot of them has similar needs. So my thinking is I can help them for flat montly fee to do all the infrastructure on AWS for them. Obviously this is no way to get rich but my hope is to streamline my process and avoid chasing invoices etc. What do you think?

Open articleView on HN

Nice idea! I've got some questions:

- do you handle any incidents that arise within aws? Say there's a problem with my ec2 instances from Amazon's side. Will you track this or notify me about it? Or do you provide the infrastructure with tf and after this I am on my own?

- what do you do when a customer changes things manually but then wants you to change something for them. Your terraform state will probably have conflicts.

- what is your workflow? I assume every client has its own git repo and tf-state? Do you manually apply the changes?

Hi, thanks for question,

- yes I'll handle indidents while you're on subscribtion. After you're off subscription you're on your own (I can't support all clients forever for free). But as I mentioned before I will introduce maintenance program for lower fee (maybe starting at $99/m but depending on the size of your infra it can go much higher) to monitor and update your infra and do any fixes.

- as you know if you use terraform or CDK you shouldn't do manual changes. I inform my clients about that and they should be able to do little changed themselves using terraform. If they still do it manually it will need to be deal with in a request. I give them choice to go with either TF or CDK whatever is more convenient to them so after subscription they can live without me. As mentioned before Maintenance/monitoring plan can deal with some of these things.

-yes every client has their own git repo and tf-state which they are owner of. Changes are applied manually for now. I was also looking into TF Cloud (HCP) but it's other cost for customers and other thing they need to know.

Thinking about it know thanks to yours (and others) questions maybe I should try/offer to keep all happy customers subscribed to some cheaper (from $99) monitoring/maintenance plan and take complete care of their infra. (Until they actually need full-time in-house person.)

Thanks for your answers!

I suggest you to think about the pricing for the monitoring/maintenance plan really good. 99$ for just monitoring and informing the customer sounds fine. For maintenance - 100$ is basically 1 hour of work, so if there's more than one incident per month that takes longer than a hour, you'll lose money.

I spotted a typo on your site. It was then I realised that you have nothing in the way of "contact us" or "who are we?" or any other way to get hold of you.

I'm not sure I'd want to give my infra keys to a team where the only way I can contact you is Trello, and I have no idea who I'm working with.

Do you give customers more details before signing up, or is the idea to remain this anonymous?

ps. The "How This Works" FAQ has the typo "udpates"

Thank you for feedback!

Regarding contact, on the website there is an option to schedule a call where we can talk so you can see I'm real person. During that I'm happy to share any information about me. I completely missed that there is no email on the website, right now it's visible on calendly page if you go to book call. I'll add it somewhere there. Thanks!

Interesting idea. If I were to challenge it, how this differs from setting up a terraform (or other automation) that will set up and change infra for me? Also that sound like you're not writing the code, but initially DevOps was all about writing the code and shipping it yourself. Perhaps a IaaS is a better wording here, but then, again, whats the difference?

Hey, thanks for challenging this. Good questions!

It doesn't differ from you setting up terraform or aws cdk yourself. If you know how to do that or have that expertise in house and have time to do that you don't need this.

I'm programmer for 21 years, I could do coding but this is focused on infrastructure itself: Setting up CI/CD, networks, instances, rds, security, monitoring etc. etc.

I know DevOps is combination of Dev(elopment) and Op(eration)s but from what I've seen in many companies it's like 90% dev and 10% ops, and not so many devs are so keen to learn ins and outs of AWS (or other clouds) and those who know that tend to be expensive.

So idea here is to help I think mostly smaller companies who don't have enough Ops work for fulltime person. I hope this could be good option for them.

So maybe DevOps is not most correct word, maybe could be Stunning Infra or Stunning Ops but I have a feeling DevOps is more catching phrase/term.

IaaS might be better wording but again I assume that not every company who want/need/must use cloud has knowlege (or time to learn) how to setup and use it properly. (just running scout suite usually reveals lot of basic mistakes).

It's basically a different take on DevOps/Infrastructure agency.

Thank you againf or feedback!

Thanks for the answer. That makes sense. I was looking at the pricing and it may be a bit misleading because you express it as a monthly payment and the FAQ says pause anytime. So in the end I can have a single request and will only pay for the time this request used. Perhaps you can add (x$/month or y$/day or hour), that will tell the customers it's more "on demand" like renting EC2.

Thaks for follow up.

If you just pause you're gonna use all your month anyway, only later. It's a commitment. I undrestand you might not have month of work at once. Important thing for me is that I get at least $4900 from each customer - month worth of work.

If you cancel (before you use whole month) and get a refund for unused days, I assume you're not happy so you'll not come back, fair. There is always some investment on my side too to understand what is needed or what is current state of infrastructure so I want to avoid people just coming for 2 days of work for few bucks. Does it make sense?

Hey, this is interesting. We've already seen UI/Web Dev as a Service coming up. (Some of them are doing a nice job). And the idea behind this sounds amazing as well. But I've got a question. In your FAQs, you mentioned that:

> "you may not always have enough work to keep them busy at all times"

Can you elaborate further? Why do you think DevOps people might not be busy most of their time?

Hey, thank you for feedback.

> Can you elaborate further? Why do you think DevOps people might not be busy most of their time?

What I've seen few times in smaller companies/dev-shops is they just don't have enough DevOps work to keep one person busy all the time. e.g. They need to setup VCP and few ec2, some buckets, RDS etc. That might be just few days of work. Then nothing for 3 weeks and then maybe they need some other things as they work on other stuff or they got other client later.

Also they might have enough full time work for just 2 months and then nothing. That would be too much hassle to find a contractor etc.

Having said that I don't think clients should pause/resume 5 times a month but it's a way for them not to waste their subscribtion if there is not enough work at once.

I also think I should add third plan for existing customer where I'd just monitor and update their infra for lower fee (without doing new features).

Hope that make sense.

Thanks again.

This is an interesting idea for a services business; it's been done a lot with apps and sites, so DevOps logically makes sense. Having a maintenance retainer like you mentioned in another thread would be a great add on and hopefully some nice residual clients. Best of luck!

Thank you! It's always good to hear from others that my ideas makes some sense.

Wow, I didn't expect so much good feedback. You guys had some really good questions and comments that got me thinking in more detail about this. There is some work I should do to better explain on the website how this works, what I can do etc.

Thank you all!

What is the scope of a "request"?

What if my "one request" is: please provision a vpc with separate public and private subnets, internet gateaway with NAT, an ec2 instance in each and an rds instance with a read replica in the private subnet.

Thank you, it's a great question. I tried to anser that at the bottom of the website in in Commonly asked question:

>This will vary but usually it’s one atomic piece of AWS infrastructure, e.g. VPC setup, Cognito user pool, set of ec2 instances, RDS setup etc. You can create tickets as big as you wish if it’s easier for you but delivery will take longer - however we’ll push partial deliveries of these bigger reqeusts every 2 business days.

To answer your specific question, I'd say this is more than one reqeust. I'd see it as 2 requests. Depending also if you have any existing infra with some terrafor or cdk or it's a new project etc.

I understand scope of one request is tricky and different people might have different expectaions. Maybe I should think about it more and have more detailed explanations/examples of "one request scope". But it's hard as each case is specific and can have nuances.

I'm open to any ideas.

Thanks!

Reply on news.ycombinator.com