drob

11 ms·

61.
by drob 10y ago
This isn't live yet, but we expect it to be across our citus cluster. The ~30% figure comes from the profiling we did on individual postgres nodes.
62.
by drob 10y ago
We have a bag of utils internally to paper over the missing JSONB functions. This was definitely a headache at first. This is mostly fixed in 9.5: http://blog.2ndquadrant.com/jsonb-and-postgresql-9-5-with-ev...
63.
by drob 10y ago
Author here. Curious what experiences y'all have had with JSONB. We're in the process of switching to a more balanced schema (mentioned in this post) and the results have been pretty good so far. Another win has been that the bett
64.
When to Avoid JSONB in a PostgreSQL Schema(blog.heapanalytics.com)
8 points by drob 10y ago|0 comments
65.
by drob 10y ago
We've hit a lot of the same fundamental limits scaling PostgreSQL at Heap. Ultimately, I think a lot of the cases cited here in which PostgreSQL is "slower" are actually cases in which it does the Right Thing to protect your
66.
by drob 10y ago
We've used both at Heap. I wouldn't expect much of a performance change from switching in either direction. JSONB has the advantage of being in a data format that lots of tools speak, and has generally been less annoying to use. B
67.
by drob 10y ago
I'm actually halfway through writing this post! I was hoping to finish it today. :)
68.
by drob 10y ago
Or if you need GiST indexing, can't wait for someone to write the bindings for JSONB, and have non-nested data, in which case hstore is probably the right choice.
69.
by drob 11y ago
> Mostly, a select group of people are getting richer, while everyone else stagnates. Actually this isn't true. Middle classes in the developed world are doing poorly relative to the richest in the developed world, but global povert
70.
Writing Postgres Extensions(big-elephants.com)
2 points by drob 11y ago|0 comments
71.
by drob 11y ago
Profiled this. Using first_value instead of the additional sub-select made this perform slower at a factor of about 2x for arrays of 1M events. Interesting, and very surprising! I'm going to dig around and see if I can't figure ou
72.
by drob 12y ago
Author here. I didn't know about first_value. Thanks for the tip! I'll play with this later today and see if it performs any better. In any case, it makes the query easier to read, so we'll probably use this unless it somehow
73.
by drob 12y ago
I'm silly -- turns out I did mention this in the post. See footnote 3. The multi-column index takes up 755 mb, vs 1026 mb for the set of three single-column indexes.
74.
by drob 12y ago
I wonder if this has something to do with differential support from different DBs. I don't think MySQL has an equivalent feature, though I think SQLite does.
75.
by drob 12y ago
Yep, looks like a minimum delay. Interesting! Agreed in general. No one-size-fits-all answers, and schematizing your data well is going to require case-by-case attention and experimentation for the foreseeable future.
76.
by drob 12y ago
In this example, a relational schema would work fine, and might have been easier to read. We use HSTORE at heap (until JSONB ships!) because we're processing event blobs with thousands of different fields, of which most are irrelevant
77.
by drob 12y ago
This is a scenario in which you do know the query set ahead of time, and there's still a considerable performance benefit to using partial indexes over any combination of multi-column indexes. The main insight is that the "high va
78.
by drob 12y ago
Separate single-column indexes can be reused more readily. In this scenario, you might have a few different types of "high value" events, and you'd want your indexes to be applicable to more than one of them. A multi-column i
79.
Speeding Up PostgreSQL with Partial Indexes(blog.heapanalytics.com)
98 points by drob 12y ago|27 comments
80.
PostgreSQL’s New LATERAL Join Type(blog.heapanalytics.com)
278 points by drob 12y ago|39 comments
81.
How to Eradicate a Disease(moreintelligentlife.com)
27 points by drob 12y ago|0 comments
82.
Hackers Access at Least 100,000 Snapchat Photos and Prepare to Leak Them(businessinsider.com)
12 points by drob 12y ago|2 comments
83.
by drob 12y ago
Sure, but that itself is interesting: context around data systems is changing so fast that even the mature ones have to innovate pretty quickly to keep up. (Or maybe especially the mature ones?) I wonder how long before postgres includes an
84.
by drob 12y ago
I did see that, although it's plausible that an analogous change would be helpful in core postgres. The readme seemed to suggest that they tried this on cstore_fdw because that was easier on which to develop, not because there was more
85.
by drob 12y ago
I'm always surprised at how much room there is for big performance improvements, even in a system as mature as PostgreSQL. Evaluating an agg node is a pretty core piece of functionality, and there's still room for a pretty big win
86.
by drob 12y ago
Very nice! That's a thoroughly earned upvote. Measure theory is super cool.
87.
by drob 12y ago
Really neat announcement. This opens up a bunch of interesting possibilities for applications that are cpu-bound once you're on SSD, but for which the delta between SSD-backed EBS and ephemeral storage doesn't change the bottlenec
88.
A Chilling Medical Trial(nytimes.com)
10 points by drob 12y ago|0 comments
89.
by drob 12y ago
Does anyone know the specific rules or statutes that underpin this argument? What are the actual rules designating how much experience a pilot/team needs with a particular airline before they're allowed to fly? It'd be intere
90.
Matrix Sketching(ternarysearch.blogspot.com)
1 points by drob 13y ago|0 comments

More ›