joneholland

20 ms·

61.
by joneholland 6y ago
My understanding is that every front end server has at least one connection (on a dedicated thread) to every other front end server. Assuming they have say, 5000 front end instances, thats 5000 file descriptors being used just for this, bef
62.
by joneholland 6y ago
Are they still using Coral and Codigo as the RPC stack?
63.
by joneholland 6y ago
Fair. What I meant to say is “hand rolling a way to have consensus on a shared piece of data” by implementing it with a naive gossip system.
64.
by joneholland 6y ago
Running out of file handles and other IO limits is embarrassing and happens at every company, but I’m surprised that AWS was not monitoring this. I’m also surprised at the general architecture of Kinesis. What appears to be their own hand r
65.
by joneholland 6y ago
My time at amazon is what firmly convinced me that the “monorepo is the only way to develop at scale” folks just hadn’t seen something like Brazil. It (and apollo) were truly a work of art and they were designed and built so long before eve
66.
by joneholland 6y ago
If I recall correctly, later games actually packed higher throughout ram into the cartridges to work around the latency of the onboard ram.
67.
by joneholland 7y ago
We use autoscaling groups and spot instances. ¯\_(ツ)_/¯
68.
by joneholland 7y ago
We are a scala shop as well, but more and more things are being built in Kotlin. I love scala but the lack of a future is downhearting.
69.
by joneholland 7y ago
Not a lot of progress in the last year.
70.
by joneholland 7y ago
Scala also has a live repl.
71.
by joneholland 8y ago
Yes! Shoot me an email.
72.
by joneholland 8y ago
I’m hiring Sr developers right across the street at Expedia. Shoot me a note: joholland at expedia.com
73.
by joneholland 8y ago
How do you add that dynamically?
74.
by joneholland 8y ago
I don’t understand the purpose. The existing proto syntax is both easy to learn and language agnostic. I’ve watched new devs pick it up in a matter of minutes.
75.
by joneholland 8y ago
I would gladly switch from MacBooks if anyone came even close to the same touchpad experience. I keep trying the latest Lenovo’s and they still have garbage touch pads.
76.
by joneholland 8y ago
Expedia is a big time scala shop, but new stuff is trending towards Kotlin.
77.
by joneholland 9y ago
One thing that surprised me about PD is that their enterprise sales team didn’t want to play ball on an all you can eat license for a flat annual price. Because of this, they are really at risk of being replaced in large companies that can
78.
by joneholland 9y ago
This is just a mashup of some visual code service and the Watson api. There’s nothing novel about ML here.
79.
by joneholland 9y ago
My wife owned a Chevy Malibu (mid 2000s) when we met and it was complete garbage from the tired design to the terrible build quality. I’ve sat in more recent American cars and they are the same way still. Even flagship cars like the mustang
80.
by joneholland 9y ago
Why would I buy a Lincoln when I can buy a BMW for the same price? Getting the product wrong is more than just the design and build quality. It’s also picking the right price point and knowing when to rebrand. Hyundai knew they couldn’t com
81.
by joneholland 9y ago
I see plenty of cars on the road. They are Toyota’s and Honda’s. I love how American companies are justifying weak sales as the economy and not their complete failure to make a compelling vehicle in the sedan space.
82.
by joneholland 9y ago
No modern company is pushing esb anymore. It’s an architectural failure still being pushed by vendors and has been architects. I use it as a canary to know if a company is technically inept.
83.
by joneholland 9y ago
Why would anyone run this themselves when ec2 parameter store w/kms is a fully managed and free solution.
84.
by joneholland 9y ago
I took his guide at the end and apparently we are supposed to do micro services, which is good, because we do as an engineering org of 800+ people.
85.
by joneholland 9y ago
You run into all kinds of hard and soft limits when you have thousands of instances. The other thing to think about is scaling for people. Do you really want 100+ teams in one account? What about cost tracking ? Etc etc. Lots of reasons to
86.
by joneholland 9y ago
When you are a large company, you outgrow the limits of a single AWS account per environment pretty quickly, and you start splitting out functional areas or services into their own accounts. These endpoints solve the last piece of the puzzl
87.
Ask HN: Is anyone using rkt?
6 points by joneholland 9y ago|0 comments
88.
by joneholland 9y ago
It seems to be a perfect fit for redis, since that only uses a single core. We run redis on R4s today and it wastes most of the cores.
89.
by joneholland 9y ago
That assumes your constructor parameters are all unique types.
90.
by joneholland 9y ago
The use of read barriers are exactly the same technique that Azul does with Zing, it’s “pauseless” GC. This allows for concurrent heap compaction, at the cost of throughput. Azul used to sell commercial “java machines”, with hardware suppor

More ›