[Osmf-talk] Community consultation: plans to hire a Senior Site Reliability Engineer

[Roland Olbricht]

Hello all,

first of all, I'm happy to comment on the job description in this stage.
Putting the job description for discussion here first before the board
meeting is a further important step forward in terms of openness.

Also, as this job description is apparently coined by Tom and Grant as
the actual sysadmins, I would like to congratulate you for asking the
right people.

> *For example, current project ideas include:*
>
>   * AWS auditing and improvements
[..]
>   * AWS

I would strongly discourage you to have any references to AWS in the job
description. If you write "cloud" then everybody will take the biggest
player in that market into account. Conversely, it is not so obvious
that people consider competitors or even non-cloud alternatives as
potential solutions if you suggest the market leader.

One core intent of Open Source is to avoid vendor lock-in. For this
reason a provider-agnostic approach is key.

Another one is a matter of money. The back-of-the envelope cost from a
unrelated to OSM for my day job:
- doing it ourselves on rented dedicated servers (three instances on at
least two sites) would cost 2000 EUR a year, and from past evidence the
relatbility is around 99,9%, although not contractually asserted
- awarding it to a local provider with a 24/7 high availability contract
would be around 20.000 EUR a year
- implementing the service on AWS with the desired availably assured had
a projected cost of 50.000 EUR a year
AWS is in a low load but 24/7 uptime scenario a relatively costly thing.

Yet it was a matter of discussion because a manager in the loop had the
attitude that he knows the buzzword AWS and has budget for it, and any
substantially cheapter solution must be evidently faulty. Once again a
reason why avoiding reference to the brand may turn an otherwise
uphill-battle for both openness and cost-efficiency.

That said I strongly encourage to prefer a candidate who firmly knowns
how to set up containers over a candidate who is entrenched with AWS.

I would like to suggest for this to
- move in the project ideas the line "Moving some infrastructure to
containers ..." to the top
- drop the "AWS" line
and
- drop "AWS" from the must section with no replacement
- promote "Containerization" to the must section

But all in all, I'm glad to see that things are getting done at an even
higher pace than before.

Best regards,

Roland