Issue #129
Deep Neural Nets - 33 yrs ago and 33 years in future, Dist sys shibboleths, Use fast algorithms, Tracing Single Operation in Dist Sys.
"remember that sometimes
it’s ok to close your eyes
and forget the world,
breathe deeply,
and do that one thing
you’ve been putting off
for months
and when you are ready
to view the world with
new-born eyes,
you will find that
the stars still shine,
the waves still crash,
the sun still rises,
and
your heart still beats."— Shelby Leigh
Deep Neural Nets: 33 years ago and 33 years from now
In 2055, you will ask a 10,000,000X-sized neural net megabrain to perform some task by speaking (or thinking) to it in English. And if you ask nicely enough, it will oblige. Yes you could train a neural net too… but why would you? - #AndrejKarpathy
Wide-column, Column-oriented and Column Family Databases – A Deep Dive with Bigtable and Cassandra
Simple explanations of Bigtable, Cassandra, Hbase - #Shivang
Distributed Systems Shibboleths
I am still a relatively new engineer in the field of distributed systems, having only studied and worked in the field for around a decade, but in that time I believe I have learned to recognize some key “distsys shibboleths” that help me recognize when I can trust what a vendor or other engineer is telling me. - #JoeyLynch
Getting comfortable with being uncomfortable
Have you felt that feeling? That moment of uncertainty, where you don’t know what the solution will look like. You’ve solved many other problems before this one (and so far they keep paying you), but now you have to reach into the creative ether again to come up with a way to solve this new one. - #AlexEllis
With a bit more observation, you realize your initial impression was entirely wrong. Gloria does indeed do nothing much of the time. But every so often, a request, instruction, or alert comes from Tony and she leaps into action. Within minutes, she answers the call, sends the letter, reschedules the appointment, or finds the right document. Any time he has a problem, she solves it right away. There’s no to-do list, no submitting a ticket, no waiting for a reply to an email for either Tony or Gloria. - # Shane Parrish
As an engineer who primarily works with data and databases I spend a lot of time moving data around, hashing it, compressing it, decompressing it and generally trying to shovel it between VMs and blob stores over TLS. I am constantly surprised by how many systems only support slow, inefficient, and expensive ways of doing these operations. - #JoeyLynch
The diminishing returns of productivity culture
We’ve reached the point of diminishing returns. I think we know this. You can see it explicitly manifest in anti-hustle culture, in the renewed embrace of unions and the labor movement, in the popularity of books like How to Do Nothing and movements like The Nap Ministry. Some people have known it for a long time, some are just gradually coming to terms with it. A lot of it, I’ve found, depends on just how inculcated you were by productivity culture. Were you surrounded with examples of productivity as success? Or were the “productive” people in your life the most exhausted and pissed off? Maybe a disability meant that productivity was never an option, and you had to learn to live in a society that venerates it above all else. Maybe you always rejected it. Maybe you’re from a country that doesn’t worship growth at all costs. - #AnneHelenPetersen
Apache Kafka Rebalance Protocol, or the magic behind your streams applications
The rebalancing protocol is an essential component of the consumption mechanism in Apache Kafka. But, it also serves as a generic protocol for coordinating group members and distributing resources among them (e.g Kafka Connect). Static Membership and Incremental Cooperative Rebalancing are both important features which provides a huge improvement to Apache Kafka by making this protocol more robust and scalable. - #FlorianHussonnois
Tracing a Single Operation in Distributed Systems
But what if we want to solve the problem of knowing why a single request is slow? It is tough to find a single request in profiler results. Additionally, as we will explore in this article, this problem becomes increasingly complex as we move from monolithic applications toward microservices, distributed architectures, and distributed systems. - #Krzysztof Slusarski