RT @sama: ok cool will you do that for retail investors on the way down too?

Stability is never perfect. Recoverability is important. When things get overloaded or crash, how fast can the problem correct itself?. Some users may not even notice the difference.

Which, as he says, is the point. It's not a tradeoff between going faster and going in the right direction. They actually complement each other.

There's still a difference between setting high expectations, and telling your team they have to work 12 hour days doing ineffective tasks or covering up for someone else's mistakes. High performance is only sustainable if it's going in the right direction!.

Snowflake CEO Frank Slootman wrote about how he creates high-performance culture 2 years ago: Interesting perspective on pushing the limits of what is possible.

If you spend too much time fixing problems that are rare, low impact, or better addressed in another way, it gets hard to do new things. It’s worth asking whether the fix is worth that cost.

Whether you’re running a business or managing a team, try advertising your staff’s competence — to them. People often don’t know the amazing things they have accomplished. Your job is to remind them!.

Great to see how Google handles impostor syndrome: Make it a part of the conversation and it's a lot less of a negative force.

Just like writing code for a feature that customers never end up using, if you spend a lot of time on processes for documentation, priorities, release management or QA that don't lead to any productive result then it's worth revisiting that process.

A lot of effort goes into creating abstract code where you can plug in interfaces to have it work with anything you want. Yet a lot of times when you do it that breaks the assumptions of the abstraction anyways. Premature optimization and all that. .

Always beware of over-engineering -- using the latest and greatest programming techniques doesn't help if your code can't meet the customer needs or internal business needs in time.

Motivation, when you're undertaking some new goal, is not just about seeing immediate results. It also depends a lot on believing that the small progress you see at first is a sign that you're going in the right direction and it will lead to something bigger.

Read replicas are a common solution to scale SQL databases. They come with a cost — your code has to support query routing and replication lag. Working on an article about a clean and simple solution to get those benefits without the cost!.

Tools can speed up what you're doing. They can also encourage you to do it the wrong way. One of the biggest rules of project management is to not take on more than you can accomplish. Yet every project management tool makes it really easy to add tasks until you end up there!.

Gave a talk on load testing and scalability for @phpvancouver last night: Thanks to Vancouver PHP Meetup for hosting! Great questions at the end. #VanPHP #PHP #aws #webdevelopment #backend #backendwebdevelopment.

Sharing my experience on how to do great load tests tonight: This is online so anyone can join!.

Development projects never get done if people keep throwing in more things that you can do "while you're at it". Focus on the value and what actually gets you there. Everything else can be done separately.

RT @richard_g: @DanAshby04 @michaelbolton @keithklain “We need a small change to the thing we released yesterday”. “That will take a month”….

Good education, in all its forms, is as important as creating something entirely new.

The cutting edge of new technology always gets attention. But there are vast amounts of incremental progress to be made in implementing what we already know. And it largely depends on helping everyone understand the trade offs even if they haven’t worked with the tech directly.