Hacker News Reader

Comment by latchkey | Hacker News Reader

Back in 2005, we ran the worlds largest hardcore porn site (kink) on JBoss 4, jgroups with multicast, EJB3 and we did all of this on 3 servers, but only needed one and used the rest of for redundancy on fully automated rolling deploys. They weren't even very special servers... just some Dell blades.

We had full transaction isolation and it just ran thousands of connections and transactions per second, gloriously, without any issues. We had a micro currency (kinks), first 1080p pay per minute live viewing, and a whole slew of happy customers with .gov/.mil email addresses.

Kids these days with their 'microservices'... psssschhhaaa...

View on HN

Yep. But I applaud it; let my competitors please adopt AWS, k8s and microservices. We run our monoliths on a few $/mo VPSs while our competitors run 1000-100000s $ per month (literally met the CTO of a UK competitor a few weeks ago at a meetup and he explained their prod + staging for their few clients is now up to $80k/mo at aws; comparing to run around $2000 in total hosting for all our clients together for staging, prod + test. Of course that competitor is playing with investor money so they don't care). Our way requires vastly less resources; humans & machines, so we run far higher margins (in our industry, pricing is identical for all including the sales discounts that are given), we sleep better and more (monoliths are so much easier/faster to debug and deploy) and when things are down (like last year) in the market, we have stacks of profits while two of the top competitors had to get new funding to not just disappear (both did get the funding).

I wish everyone many microservices and cutting edge frontend frameworks!

Less cost, sure. Faster to debug? Eh, possibly, in some situations. Faster to deploy though... How could that possibly be true?

One binary on a vps? To deploy all we have. How can it be faster/easier?

Do you have zero downtime upgrades and automated rollback on deployment failure?

This times 1000. Devs way over estimate their workloads and way under estimate the ability to scale with fairly low effort. Creating distributed spaghetti is still spaghetti - but now your bowls are spread all over the map.

Reply on news.ycombinator.com