Feeddo City – Stage 2

Part of the interest of building a service is the learning curve of designing, scaling out and monitoring the hardware layer. It is good to see Feeddo reach to stage 2. So far, so good! (And thank you to all the passionate and hard working people developing amazing open source products!)

Feeddo City - Stage 2

(Thanks Oliv!)

Author: @feedly

Read more. Know more.

2 thoughts on “Feeddo City – Stage 2”

  1. Edwin,
    It would be good if you could share your thoughts on why you chose to build vs. rent. Did you consider using a traditional hosting or utility service?

    Anshu

  2. Hi Anshu,

    When we started, we did not have enough experience either way to be able to make an intelligent decision so stage 1 was a single server hosting all the back end components.

    How that we have a better understanding of the parameters of the problem (CPU, security, i/o, RAM, Amp, disk, back up, etc…) of each of our back end services, we are scaling the initial stage 1 set up to 3 nodes (20 cores, 20GB RAM) – aka stage 2, aka “block”.

    A block should be able to support 250,000 active users (with a 50% error margin). Our goal is to use this setup to continue to learn and validate the model and fine tune things.

    Stage 3 will be about moving stage 2 to the cloud and replicate blocks to scale horizontally.

    The top 2 lessons for me during the stage 1 + stage 2 were: …1) how important is the number of amp that your machine consumes under load because it directly correlates to the co-location price and …2) the effort/knowledge it takes to secure the machines.

    A good learning experience overall. Very much like planning a city…hence the feeddo city term.

    Will continue to report on our learning experience.

    -Edwin

Comments are closed.