Medijski sadržaj
- Tweetovi
- Tweetovi i odgovori
- Medijski sadržaj, trenutna stranica.
-
Almost everything I see/hear about huge pages in Linux lend credence to why it's probably never to be used. Is there any particular scenario where it actually helps? Genuine question.pic.twitter.com/qRVpiqc81O
-
Tweet brought to you because I’m reading some chapters from the new book on how software engineering is done at Google (think of it as the SWE version of the SRE book, each chapter authored by someone different) and the first thing it talks about is java builds lol.pic.twitter.com/901u4Dn133
Prikaži ovu nit -
Writing up my “best of 2019 in tech talks” post and somehow not one but two talks on ... derivatives of radix trees have ended up on the list. 1. Maple tree, used in the Linux kernel 2. EBTree, used in the HAProxy task scheduler I personally
tree data structures.pic.twitter.com/Cx1B5HcSZ2
Prikaži ovu nit -
I’m writing up my “best of 2019 in tech talks” post. one of the talks that made it was one which said most “AIOps” is snake oil. you need massive amounts of data to train models to “detect” incidents, which isn’t feasible without having millions of incidents a day.
https://twitter.com/ReinH/status/1199413250078560256 …pic.twitter.com/ebzMBivR0r
-


this talk by @ifesdjeen on transient replication with witness replicas and cheap quorums in Cassandra. Is definitely going in my Top of 2019 list! A lot of this is derived from a *very old* paper (Voting with Witnesses) I tweeted about 2 years ago https://twitter.com/copyconstruct/status/1015854648622604288?s=20 …pic.twitter.com/8wzsfAdZsY
-
The Bay Area handles rain really badly. A huge puddle on every street corner with parts of Lower Haight looking downright flooded. Wet socks are the worst ever. Thankfully I had a pair of spare socks at work. I’ve never felt more cosy (or looked more ridiculous).pic.twitter.com/kid4NYrrWY
-
There's also a great talk by the authors of this post at SRECon 2019. https://www.youtube.com/watch?v=ozuo4WPOH9g … Where they talk more on the failure modes, "ring resizing", and Zookeeper (heh). PS: If you use Finagle, you get this for free, and the implementation is an easy read.
Prikaži ovu nit -
This is such a great read. Touches on so many topics: - installing control loops for feedback (PID loops-esque?) - cons of statistical load distribution - the limitations of this approach - eventually consistency is the way to go (even when coordination is required *ahem* ZK)pic.twitter.com/5XWCYUdsnF
Prikaži ovu nit -
This post on the evolution of client side (distributed w/o coordination) load balancing at Twitter is
- [service] "mesh" topology + P2C has cons
- the "random aperture" approach which scales unfairly
- "deterministic random aperture" scalable + fair
https://bit.ly/2FvTH2u pic.twitter.com/D4JN9jjahD
Prikaži ovu nit -
Deterministic simulation didn't exactly become ubiquitous the past decade (not enough thought leaders telling people they're doing it wrong if they don't use simulation), and FDB's transactional guarantees has caveats. Still,
@rbranson and co are doing some cool work with FDB!pic.twitter.com/KFuQ19KQTr
-
I’ve been working on this post since Nov 2018 and it’s nowhere close to done. I lost spring of 2019 to allergies and then got super busy with a bunch of other things. This is the first post in my drafts that I want to be done with this by this spring (if I manage to find time).pic.twitter.com/2JL61f4xXP
Prikaži ovu nit -
Don’t know how I ended up a rabbit hole where I’m now researching things like cache partitioning, register allocation via graph coloring, register renaming etc, but wanted to back up a bit and brush up on what I remember from computer architecture 101 from college a decade ago ..pic.twitter.com/jQ5wdpcXUa
Prikaži ovu nit -
The Spectrum of Compute This is not an exhaustive comparison, but what with bare metal servers, VMs (including lightweight options like Firecrackeror or alternatives like gVisor), containers, serverless, ‘edge compute’, dedicated hardware like TPUs, options galore in the cloud.pic.twitter.com/gzRLR9Pa1j
Prikaži ovu nit -
Beyond the hype ... breaking down Kubernetes my tongue is firmly in cheek and this isn’t meant to be taken too seriously ... ... expect maybe the bit about k8s in its current form as a set of abstractions for building a PaaS feeling super dated already in 2019. That’s true.pic.twitter.com/IPywGAahSZ
-
Was chatting with a friend today about tech marketing, and ended up using Docker as an example to break down how it might help to have some kind of framework to understand what problem a piece of technology is solving. Feel all new tech should be analyzed in the same way.pic.twitter.com/3ipxCeofQd
Prikaži ovu nit -
This article from the Amazon Builder's Library might be on the evolution of continuous deployment at AWS, including pre-production testing, config testing and testing in prod ... ... but the most important takeaways here are actually the cultural ones. https://bit.ly/34A2yKr pic.twitter.com/wYH1JDF3Sp
-
Japan was a stationery and craft store lover’s dream. I spent hours inside stores in Tokyo, even if all I bought was a few pens and some craft paper. They have 6-7 stories buildings that just sell stationery.pic.twitter.com/RrGEhFypi8
-
Started reading the Amazon Builder's Library today. The very first article I read on deployments was absolutely amazing! It goes into great depth about the challenges of deploying *stateful* services, in a manner I've never seen it written about before. https://amzn.to/2suymTz pic.twitter.com/L7VYP9Wt14
Prikaži ovu nit -
Continuous integration and deployment with Bazel at Dropbox Interesting article, which explains how despite using a monorepo, Dropbox manages microservices-esque deployments. I wonder how other orgs using monorepos do this. https://blogs.dropbox.com/tech/2019/12/continuous-integration-and-deployment-with-bazel/ …pic.twitter.com/l1MD0OnGJk
Prikaži ovu nit -
"Rather than acting as a server replacement, Lambda is effectively a force multiplier, with each CPU core overseeing the work of 100s of others. " "roughly double the perf by avoiding spurious allocation in Go" https://www.honeycomb.io/blog/secondary-storage-to-just-storage/ … Pretty awesome stuff from
@honeycombiopic.twitter.com/08Crtd9OT7
Čini se da učitavanje traje već neko vrijeme.
Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.