HNHacker News
TopNewBestAskShowJobs

ADefenestrator

18 karma · joined December 21, 2016

submissionscomments
ADefenestrator··on Java’s new garbage collector promises low pause times on multi-terabyte heaps
It's not 100% wrong. "Everything is an object" comes with some extra memory overhead, plus some GC accounting. Most garbage collectors need a decent amount of "slack" space to avoid OOM or long STW collections as well. So, I could see Java needing 1.5-4x the memory of C++ pretty easily anyways.
ADefenestrator··on Java’s new garbage collector promises low pause times on multi-terabyte heaps
Yeah, CMS is great until it isn't, and then it ruins your whole day. A bit of a black art to get tuned well, too, and when your workload changes slightly your tuning is no longer effective and the world stops for 30+ seconds. Hopefully ZGC will be more like G1 in terms of "feed it some extra memory and set a latency target" being all the tuning you need most of the time.
ADefenestrator··on Java’s new garbage collector promises low pause times on multi-terabyte heaps
Oof. There's probably some compounding factors that make it more difficult, but usually getting GC pauses <500ms isn't that hard with current-gen collectors. Either being very careful with CMS, or just feeding G1 a lot of memory.
ADefenestrator··on Open-sourcing a 10x reduction in Apache Cassandra tail latency
It's still an issue, but a lot less of one. 3.0 is a big improvement in terms of GC behavior. Haven't tested 3.11.x yet, but it look in theory like a decent improvement in terms of rounding off corner cases and adding instrumentation.
ADefenestrator··on Open-sourcing a 10x reduction in Apache Cassandra tail latency
FB abandoned Cassandra (which was really only used for message inbox indexing) when they redid how messages work years ago, but the re-adopted a large C* infra when they bought Instagram.
ADefenestrator··on Open-sourcing a 10x reduction in Apache Cassandra tail latency
For a small to medium sized shop, sure. For someplace with thousands or tens of thousands of nodes, the new system ends up cheaper in the long run.
ADefenestrator··on Open-sourcing a 10x reduction in Apache Cassandra tail latency
The good news is, at least for my workloads, garbage collection is significantly better in 3.0 than 2.1 even without off-heap objects. Not sure if it's generating less or just generating it in a way that's easier to collect cheaply, but I saw pause times and total collection work drop significantly with the same settings (G1 collector)
ADefenestrator··on Ask HN: If you've used a graph database, would you use it again?
TAO is probably better described as a custom graph cache. It understands the data model and can do certain operations (especially stuff like set intersections), but the authoritative store for the data's still MySQL
ADefenestrator··on OVH outage explained
Easier to generate a fixed-frequency AC output. Diesel engines can run at a fixed RPM and adjust fueling to match load, where gasoline and natural gas need a narrow AFR and have to vary rotation speed for power output. Maybe less of an issue these days as inverter/generators are more common, though I dunno if that's the case for the big generators.
ADefenestrator··on OVH outage explained
I thought large diesel generators generally did operate at a fixed speed in order to output 60Hz. I know inverter/generators are common these days for small gas gensets, but have the big diesel ones switched over as well?
ADefenestrator··on Why Did So Many Startups Choose MongoDB?
A lot broken-ness is still left for the sake of compatibility (especially since HHVM's open-source), but Collections and type hinting make a huge difference.
ADefenestrator··on Why Did So Many Startups Choose MongoDB?
These days MongoDB more or less (finally) works as advertised as well. It actually has worthwhile backend options (plural!) and the new (v1 vs v0) replication protocol is actually correct. Eight or so years of development has led to the most egregious of the bugs and corner cases being rounded off.

So, it started as a shitshow on the back-end and people who had to maintain systems based on it hated it, but it got popular because it was easy to get simple systems up and running with. Then, over time, it slowly became less awful as it was substantially rewritten and large portions of it redesigned. I'd say PHP is a very, very good comparison.

ADefenestrator··on Is it possible to host Facebook on AWS?
I think working for Akamai you're looking at things in a rather CDN-centric sort of way, but shuffling bits around is a relatively small percentage of what Facebook does. Actually doing things with the data beyond serving it up takes a lot more servers. The "CDN-equivalent" servers in PoPs etc for FB are a a pretty small percentage of the total, and the bulk are things like web/cache/search/DB servers.
ADefenestrator··on Designing 100G optical connections
It's coming from loosened tolerances. Higher/lower temperatures may mean higher loss. I'm assuming there's a bit more tolerance in the alignment of the up optics/connectors as well. So, if you get lucky it might well be 5db@2km, but it's not guaranteed/required by spec.
ADefenestrator··on Banks scramble to fix old systems as IT 'cowboys' ride into sunset
Right/left Democrat/Republican is also a bit different at the local level around those states. There's the moderate conservatives, AKA Democrats, and the hardline conservatives, Republicans.
ADefenestrator··on Banks scramble to fix old systems as IT 'cowboys' ride into sunset
Yes and no. Some of the more successful "startups" like Google and Facebook have slowly rewritten the bulk of their systems over time. In the case of Facebook in particular, there's very little if any of the early scary spaghetti PHP left.
ADefenestrator··on Building a scalable time-series database on PostgreSQL
I absolutely agree about hipster tech and databases, but I'm not so sure that Cassandra fits entirely into the "just works" category. Don't get me wrong, it's pretty neat and definitely more mature than stuff like Mongo or whatever other NoSQL-of-the-week, but it's not exactly at the level of PostgreSQL or even InnoDB/MySQL. It works, but it needs quite a bit of care and feeding by comparison.
← PreviousPage 2 of 2