• Randal L. Schwartz (github)

    Frlast!

  • LXE (unregistered)

    How's that a WTF? The architect was rewarded and promoted, and Azure got a bigger check, which I (as a former Microsoftian) wholeheartedly welcome.

  • (nodebb) in reply to LXE

    "Everything is always for the better."

    Just not necessarily for YOUR better :)

  • Young Engineer (unregistered)

    I've never related more to a post before.

  • MRAB (unregistered) in reply to Randal L. Schwartz

    If you want it to operate at webscale, it should be Frlast! Frlast! Frlast! Frlast! Frlast! Frlast! Frlast! Frlast! Frlast! Frlast! Frlast! Frlast! Frlast!

  • Aspie (unregistered)

    A group of ex-colleagues got together and the concensus was that the original system, scaled to 3x peak load, 20 years later, would be fit to cope with 2.75x peak load. The cost and size of the architecture today is orders of magnitude greater. We've gained all the downsides of a distributed architecture. Not sure if the business has gained anything. Most techies have gained great CVs though

  • DM (unregistered)

    Agree, how is that a WTF? Only 75% more spend, What kind of figures of related spend for managing what % of turnover on what total company turnover?

    No write up on any points that improved nor direct soundbite from the lead designer. Just one sided whinging to probably a little biased.

    Maybe it was 50k a month cloud spend, now it's 87k5 a month, on a 100 employee business, managing 95% of the total 35mil turnover? That would not be a WTF manuvre.

  • Donnie (unregistered)

    You're describing my workplace a year ago. We had an Alvin, and he was famous for the exact same mumbo jumbo, empty word salads, and defense of approaches that were unmaintainable. Data replication like you described, actually didn't work, because developers would screw it up, or the schema would change. We were having tons of discrepancies thanks to Alvin. Our infrastructure costs are 7 figures per year, yet this infrastructure is struggling to perform as well and as reliable as an SQLite database on a raspberry Pi, with a Flask back end. And I'm not exaggerating one single bit. I'd have to pay a professional saboteur and they'd still do a worse job than our Alvin.

  • (nodebb) in reply to LXE

    Not to mention everyone who worked on the project got a few more bullet points for their résumé which is about to come in handy.

  • JustaDBA (unregistered)

    Wrapped up the 2020s in a nutshell. Love it.

  • King (unregistered)

    No way that the Azure bill was only 75% more. More like 750% more.

  • hhja (unregistered)

    Well in mu compsny everybody is Trzinu to convince ne that we need kibernetski. And our scalibg problema Will be solved. All i'm saying is that the problemm is scalibg which is od highly spiky nature on memory front. But quite tepeatwble on bi weekly basist. And that evertyhig we need is to store the memory per minute per day for a month. And instead od using auto scaler use the scaler api to scale the servers. Up front

  • (nodebb)

    It seems the senior architect Alvin has found this article and added his perspective. :)

  • Richard Brantley (unregistered)

    The spend was not 75% more, it will absolutely be more. Imagine the nightmare it will be to add new features or change the behavior of the system; this sounds like a scenario where a small change that could be done in a week or two will now require a month or two and pray it doesn't break something. Multiply that by all the changes the user base will ask for as the business changes and velocity will will decrease rapidly.

  • (nodebb)

    "it could cost them a few thousand a month to operate... Which meant they were spending hundreds of thousands a month to host the application.... they had to spent almost 75% more on cloud services."

    Which is it, 10 times as much, or 75% more?

  • (nodebb)

    According to the vid "Don't build a Distributed Monolith: How to avoid doing microservices completely wrong", having the (relevant) admin data replicated across all databases that use it is the way it's supposed to go... but certainly not by manually updating all of them for each change.

    You're supposed to propagate the changes through an event bus, or I guess automatic replication at the database level can work too, as long as you keep in mind that updates are not supposed to be instant.

    Oh, and microservices actually calling each other is a big no-no.

  • online calculator (unregistered)

    calculate online islamic inheritance with inheritance calculator.Distribution laws to ensure fair and legal wealth distribution under islamic principles https://islamic-inheritance.mohammadharoon.com/en/

Leave a comment on “Best of 2023: The Microservice Migration”

Log In or post as a guest

Replying to comment #:

« Return to Article