Read / Write Ratio

Ibrahim, why don't you write more?

In distributed systems, you need to carefully optimize for reads or writes based on the expected usage of the system.

Is there a lot of updates, and data coming in with only sporadic reads, or mostly stables content with lots and lots of viewers?

• Database tables can be normalized or denormalized to for faster reads or faster writes. Not having to join makes things faster, but it takes extra time and disk space to prepare the materialized view.
• Compared to S3, Glacier is there for data you want to keep for long periods but might not need to read unless required - like old medical/financial records
• In Search, you can tokenize more at indexing time (writes) to make queryings (reads) faster, or you have more complex queries and simpler indexing

My personal ratio is dreadful. I read a lot of twitter, books, blogs, etc, but have only written a couple of blog posts.

From my personal perspective, I'm high read, low write, but from an external perspective, I'm high write (into my brain), and low read (from my brain).

As systems are balanced to optimize for one or another, the following hacks improve my write throughput (read from brain throughput)

Step 1: Reduce reading

sudo hostsman -i twitter.com:0.0.0.0

Blocking access to some sites, etc. Having a smaller phone. It's working, but I need something more aggressive like

while True:
    sleep(random(20 * 60 * 1000))
    for tab in chrome.tabs:
        if tab.url.match(/twitter|facebook|.../):
            tab.close()

Step 2: Increase Writing

The biggest earlier challenge was thinking too much. Lowering the bar to OK is Enough, and focusing on incremental value helps a ton.

What works now is supportive readers, and having fun writing

exec(✍️)
print(🙏)
max(🥳)

Improving read from brain throughput - something good has got to be in there!