This message was deleted.

It's difficult!! I've done it wrong a few ways and more goth in others. I think you can be evolutionary- which builds a culture of engineering over a long time (2+ years) or revolutionary (much shorter which gets you results quicker but upsets the apple cart) so you have to be tightly aligned with the VP, C suite and choose an approach or combination that works into your company's strategy. I've been a part of quite a few and been a leader for four or five now and there's no clear cut way that works for every situation. • Know your engineering and company culture • Set achievable and measurable goals • Set expectations and be realistic not optimistic

Great question @Linda Shay . I think @Schuyler Bishop makes some excellent points. I’d add two thoughts: firstly, context is king, so consider everything said in the talks may not apply to your domain. Second: my first step in the process you describe would be to appoint a product manager. The first steps on the journey towards a great platform should be about transforming to a product-centric approach.

Completely agree with the product manager. It’s an absolute key position to get right!

Hey everyone, I find this discussion really interesting, so I would like to add my 2 cents here. First of all great point about having a product manager, but having a good product manager for a platform engineering team is extremely difficult as he/she has to be good both technically and from the product side as well which is a very rare combination. In my opinion if someone is starting with a platform engg team then they should first focus on their solution/product with the existing team and from their see what all skills/requirements they need.

That's what we did. It started with some core infrastructure team. All of the PM’s (grew to multiple teams) were senior engineers from infra or product teams. The chief architect/head PM was a well known open source product developer with several startups under his belt.

I think one big issue is going to be skill sets. Not everyone who is doing Ops will want to build tools/services. Some won't have the skills or even the desire to spend a lot of time writing code. So assessing that is a big part of answering the question of whether to hire new folks for those roles. Having a division between platform folks and Ops/SRE makes sense to me, but it depends a lot on your shop and culture.

Funny - autocorrect changed

right

to

goth

in my initial message ¯\_(ツ)_/¯ I wish autocorrect was better in context! Poor UX! 🤣

+1 one on PM/TPM - a platform is a product whereas most DevOps teams are service (ticket)-oriented.

“DevOps team” is definitely an anti-pattern in every context.

Someone even created an anti-pattern description for it 😂 https://web.devopstopologies.com/#anti-type-b

Indeed they did! I used that from your website for an internal Walmart talk years before the book was released. You guys have been VERY helpful to me over the years. Thanks!!

Thanks for spreading the word!