Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
russss
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
16 ms
·
151.
▲
by
russss
17y ago
Fair point, but most DBs (including MySQL and Postgres) don't cluster their PKs by default - those posts are both referring to SQL Server which apparently does (and that's seemingly an implementation detail).
152.
▲
by
russss
17y ago
We do deny it. http://www.last.fm/forum/21717/_/535934/_/9525592
153.
▲
by
russss
17y ago
CBS is not part of the RIAA ( http://www.riaa.com/aboutus.php?content_selector=aboutus_mem... ) and has not had a significant music recording arm since it sold CBS Records/Columbia to Sony in 1988.
154.
▲
by
russss
17y ago
I understand why my previous post would have seemed non-committal (especially when you pull out certain soundbites), so I've denied it again in stronger terms: http://www.last.fm/forum/21717/_/535934/_/9525592
155.
▲
by
russss
17y ago
Sun have been courting us for a while, and about a year ago they managed to get our custom for two fairly sizeable (but still commodity x86_64) pieces of hardware to run our global DB on. Naturally, they were keen to have us running Solaris
156.
▲
by
russss
17y ago
In my experience, any differences between simple key/value lookups from Memcache and those from cache in a decent DB are easily lost in the network round-trip noise (i.e. solidly sub-millisecond). (of course more complex queries will be slo
157.
▲
by
russss
17y ago
He's getting it wrong (unsurprisingly). If you just use Memcache to cache the same results you get out of a DB, of course you might as well just throw that RAM at the DB. With any sane DBMS this will result in the same performance improveme
158.
▲
by
russss
17y ago
No, it's not an entire dump. 23andMe only decode certain bits of your DNA which are known to be interesting - they call this "genotyping" as opposed to "sequencing", which is decoding the entire genome.
159.
▲
by
russss
18y ago
Right, sorry. But my point still stands. SSD != RAM.
160.
▲
by
russss
18y ago
That's quite interesting - tiered storage for Memcache - but I already hate them for not open-sourcing it. Facebook made some amazing performance changes to Memcache and had no qualms about open-sourcing it. I'm not that convinced about t
161.
▲
by
russss
18y ago
It's $45k for 512GB of disk and 64GB of RAM. And however fast flash is, it's not as fast as RAM by a fair margin. Memcache is all about RAM. And if I had a choice of 640GB of Memcache in one box, or in 8 boxes, I'd chose the 8 boxes. You
162.
▲
by
russss
18y ago
RamSan: http://www.superssd.com/ Nice machines if you need absolutely the best (persistent) IO performance. I believe EVE Online uses a ton of them to store their game state on.
163.
▲
by
russss
18y ago
This totally misses the point of Memcache. Firstly, why do you need a disk at all, let alone 512GB of flash? Secondly, the whole point of Memcache is that you can distribute it across multiple boxes. Facebook, the largest Memcache user in
164.
▲
Operations Monitoring, Retro Style
(russ.garrett.co.uk)
25 points
by
russss
18y ago
|
0 comments
165.
▲
by
russss
18y ago
If anyone in London (or elsewhere in the UK) is wondering why the US and Germany get all the fun, we're working on setting up hacker spaces in London and Birmingham: http://hackspace.org.uk/ If you're somewhere else in the UK and want to
166.
▲
by
russss
18y ago
The most valuable cards are the supervisor modules, which are the ones which actually run the IOS. But I wouldn't be surprised if the Cisco licensing agreements apply to all their cards. All the cards have some pretty clever distributed for
167.
▲
by
russss
18y ago
Believe me, I was surprised as you are :). (I wrote the article)
168.
▲
by
russss
18y ago
Most of our hardcore data is in non-relational sources (including Hadoop). Our global DB is Postgres - that holds everything which we haven't had a reason to move off it - it's a pair of Sun boxes with 128GB of RAM using Postgres warm stand
169.
▲
by
russss
18y ago
You've got the wrong end of the stick here a bit. Postgres has had a "warm standby" system since version 8.1 which you can use to keep a second server in recovery mode waiting to go live. So as far as HA goes, that's pretty good - we use it
170.
▲
by
russss
18y ago
What I care about are the limits of products which are ready for prime time, or that we can easily get to that point. I haven't heard much about H-Store but it doesn't look like it's ready for production use.