PharkMillups

jdrowell # do I need lots of RAM to be performant? the innodb tuning wiki mentions gigs of RAM. I planned to deploy riak on small slices (like 512MB RAM)

bingeldac # that would depend on the amount of keys more than anything
512 is a little small, but if you don't have a ton of data it should perform ok
though VMs have slow I/O

bingeldac # and that will slow you down regardless of the amount of data you have

jdrowell # k. I have lots of keys. does riak keep them in RAM (like redis does)?

PharkMillups

BBHoss # are mapreduce queries intended to be live or batch? When I get upwards of 10,000 things in a bucket, it takes too long to run the query, like 10 seconds

drev1 # BBHoss: bucket based queries are not recommended in production. the queries work best against a known list of bucket/key pairs

BBHoss # drev1: is that a new feature?

drev1 # which?

BBHoss # querying over bucket-key pairs

PharkMillups

florian # hi guys

seancribbs # howdy

florian # i'm an engineer at Twitter and have started to look into Riak a little bit

benblack # florian: do you work with rk?

florian # what's the largest cluster that you guys now that's currently running?

PharkMillups

misaka # Hiyas. Have a Q about riak ...How good durable is it in a single node setup? If I use Innostore, can I assume it'll be as durable as, say, MySQL? s/good // Everything I read talks about distributed durability, which makes sense, but my app is going to start off with only one node and I really don't want to lose any data.

seancribbs # misaka: we err on the side of writing to disk. So unless you use a memory-based backend, your data is written reliably. what you don't get in a single-node setup is availability/fault-tolerance since there's only one

misaka # Hey Sean. Thanks for that. That's what I was hoping for.

PharkMillups

freshtonic # anyone know where I can find an example of Riak post-commit hook?

seancribbs # freshtonic: as soon as we have one, it'll be on the wiki. what do you want your hook to do?

freshtonic # seancribbs: when I put something in bucket A, I want to create another record in bucket B within the commit hook.

seancribbs # freshtonic: any more specificity than that?

freshtonic # bucket B is a log of change events. Whereas bucket A contains the 'current' document, bucket B contains the history of changes.

PharkMillups

crucially # can I do conflict resolution in post commit hooks?

seancribbs # crucially: i don't think so

crucially # pre commit hooks? i guess I could from erlang

crucially # if you set returnbody and I set the content type to multipart and 
there is a conflict, i am supposed to get all siblings?

justinsheehy # you can do some conflict resolution, but you can't easily guarantee 

PharkMillups

ouvasam # Hello Using ripple is there something to delete an embedded document
seancribbs # ouvasm: one or many?

ouvasam # many


]seancribbs # the association proxies an array, so you should be able 
to do doc.embedded_docs.delete(embedded_doc)

ouvasam # thanks, but using a html form, how can i retrieve the embedded doc ? 

PharkMillups

lypanov # Is there any good reason to use Cassandra rather than riak? How does the 
raw speed compare? Im having problems finding benchmarks :)

seancribbs # lypanov: it's hard to compare them side-by-side at least, in terms of performance
their access patterns are very different

lypanov # hey seancribbs. just listened to your awesome changelog podcast. thusly my being here :)

seancribbs # cool I didn't say much in that one, at least compared to andy

PharkMillups

crucially # in bitcask if you do a HEAD request does it need to find the 
object on disk and read it in?

seancribbs # crucially: with any of the backends, yes. there's no separation of metadata from object

crucially # ok

seancribbs # so the only savings you get is over the wire to your application

crucially # so storing riakfs metadata in another bucket and store that bucket on SSD

PharkMillups

lypanov # how can one emulate a range query with riak?
another table with links? is that efficient with mapred?
wondering how to do a new uncached mapred on a historic data set efficiently

benblack # going to be a tough one
doc stores are good at a lot of things, but time series is not really in the sweet spot

lypanov # seems to be the ideal use case for cassandra afaict :(

benblack # column stores are better at it (and not as good at a lot of other things)