Hacker News new | ask | show | jobs
by stillicidious 1671 days ago
> So why is cost optimization a constant point of conversation with AWS if it's so easy?

Limited IO bandwidth in middle and upper management alongside difficult schedules (we covered that one already). Take 2 steps above engineer on an org chart and detail becomes invisible, the vast majority of tasks begin to resemble a teenager at a mall with their dad's credit card. Meaningful technical validation phases are almost unheard of in many organizations, and largely antithetical to agile.

> Why do outfits like Digital Ocean advertise on the notion of clear billing as a positive differentiator compared to AWS?

Because they market to folk who never take the time to model comparative costs. In any project I considered them for (3 I think, <$15k/year each), Digital Ocean was significantly more expensive than AWS. I follow my spreadsheets, the industry follows marketing.

> Nobody thinks of AWS as a security company

They're the only vendor I deal with who are on first name terms with the NSA and sell in tremendous quantities to the US government. CloudFlare on the other hand, to this day, default to MITMing SSL connections for new accounts and downgrading them to cleartext en route to the back end. It seems our perceptions differ wildly.

2 comments

> Because they market to folk who never take the time to model comparative costs. In any project I considered them for (3 I think, <$15k/year each), Digital Ocean was significantly more expensive than AWS. I follow my spreadsheets, the industry follows marketing.

You make valid points, but what you're missing is the same accusations you level at Cloudflare were once leveled at AWS (which is why AWS had virtually no credible competition from Y!, MSFT, GOOG for 7 years!).

Also, Cloudflare's moat isn't 0ms cold-starts, but their persistence in commoditising bandwidth. Think Amazon Prime free 2-day shipping and how that worked out...

> Limited IO bandwidth in middle and upper management alongside difficult schedules (we covered that one already). Take 2 steps above engineer on an org chart and detail becomes invisible, the vast majority of tasks begin to resemble a teenager at a mall with their dad's credit card. Meaningful technical validation phases are almost unheard of in many organizations, and largely antithetical to agile.

I'm sorry, but what are you trying to say? "Antithetical to agile" - is there a point beyond some non-nonsensical, made up scenario? Every major organization with a cloud budget cares about optimization today at some level. This hasn't changed in over 20 years because those budget dollars used to be directed at data center costs. Now they're more fluid and can be more impactful when people make mistakes or aren't making sure to optimize up front.

> Because they market to folk who never take the time to model comparative costs. In any project I considered them for (3 I think, <$15k/year each), Digital Ocean was significantly more expensive than AWS. I follow my spreadsheets, the industry follows marketing.

Then share some examples from said "spreadsheets". Because for straight instance pricing DO beats AWS pricing in most every way. This is one of a handful of bread and butter services AWS offers (lift and shift compute). DO also does, most often, better with respect to performance (CPU/compute) when comparing directly [0][1][2][3]. This calculator [4] at DO showcases cost comparisons across all major cloud vendors and I've validated comparison in the last month with AWS - the prices check out.

> They're the only vendor I deal with who are on first name terms with the NSA and sell in tremendous quantities to the US government.

This is a rather naive comment. I happen to work in the security industry and every cloud vendor has direct ties to 3 letter agencies - that's not at all unique to AWS. Sorry to burst your bubble, but also every notable security player in the industry has similar relationships. It's not unique. It's also more advantageous for the three letter agencies in many of those relationships. Also, AWS, along with everyone else - has relationships in security information sharing for verticals. So, yes, AWS is part of FS-ISAC, as one example. Just. Like. Everyone. Else.

[0] https://www.vpsbenchmarks.com/compare/docean_vs_ec2 [1] https://www.bunnyshell.com/blog/aws-google-cloud-azure-digit... [2] https://www.digitalocean.com/resources/cloud-performance-rep... [3] https://www.upguard.com/blog/digitalocean-vs-aws [4] https://www.digitalocean.com/pricing/calculator/

> what are you trying to say

I'm saying this is my bread and butter, and I've seen the same pattern in every company I've crossed the doors of. It's no made up scenario when a team of contractors have spent 12 months building out some service, not a single person will have given a damn about costs, so cost optimization is purchased separately. The contractors aren't wrong nor is the business wrong, nor is this pattern unique to cloud spend.

> I happen to work in the security industry

Then surely you will understand how an entire subindustry can exist to mop up after engineers who could otherwise 'simply' avoid most mistakes if they just spent more time Googling.

> Then surely you will understand how an entire subindustry can exist to mop up after engineers who could otherwise 'simply' avoid most mistakes if they just spent more time Googling.

Yes, I'm sure an entire $50 billion dollar industry comes down to, like all your arguments, just lazy people and the simple fix is "Googling".