jamwt

19 ms·

181.
"If I miss a shot, so what?"(espn.go.com)
1 points by jamwt 15y ago|0 comments
182.
by jamwt 15y ago
We (bu.mp) use a lot of SSDs at our datacenter.. we've probably used ~100 64GB x-25e, and recently we have added 20+ Micro P300 disks. The first thing we used to do is try to convince the hardware raid controller not to do anything clever,
183.
by jamwt 15y ago
We are using Haskell + STM for scaling a major multicore system at Bump, and it's been exceedingly awesome thus far. However, STM happens to suit Haskell's type system-enforced purity particularly well; it's hard to see this being anything
184.
by jamwt 15y ago
Bump is hiring: http://bu.mp/jobs Come work here and help us get every man, woman, and child bumping. Or just stop by and have a beer: http://bumpinterns.posterous.com/bump-for-a-beer
185.
by jamwt 15y ago
I'd just like to go on the record to say, chances are that what we're all benchmarking here (stunnel/stud/nginx) is essentially openssl--and that, unsurprisingly, when we use whatever contortions are necessary to get OpenSSL set up in ident
186.
by jamwt 15y ago
From stud -h: Encryption Methods: --tls (TLSv1, default) --ssl (SSLv3) -c CIPHER_SUITE (set allowed ciphers) Edit: Though, you'd need to set DHE params, as another
187.
by jamwt 15y ago
Which is to say indirectly that I think the right answer is for nginx (and daemons generally) to support the PROXY protocol, or some other agreed-upon standard for a naive upstream proxy to indicate host/port information.
188.
by jamwt 15y ago
I'm reading over Matt's work carefully, but my initial inclination is not to merge the bulk of this into stud mainline. I'd rather keep stud simple and protocol-naive and have HAProxy do the HTTP work.
189.
by jamwt 15y ago
Just wait until that red box on that gray bar hovers over their every search. Their email. Maps. Maybe calendar. It's like we've all forgotten who the 800 lb gorilla is! Google's launchpad is tree-mendous. The show is just getting started.
190.
by jamwt 15y ago
Yep--low margin but ubiquitous commodity provider vs. high margin but low market share premium offering.
191.
by jamwt 15y ago
I'd love to, but there are things politically complicated about all this in the very small startup world. So maybe one day, but I can't do it right now.
192.
by jamwt 15y ago
Yep--and just to reassure you, it's more than an idea on my end... I designed and maintained a critical system a few years ago on a large K/V store that used this method, and it worked without a hitch. We never even needed to think about t
193.
by jamwt 15y ago
Kind of a side topic, but lazy, JIT migrations work well for this. You keep a "migration chain" in your fetch layer that goes v1 -> v2, v2 -> v3, and every persisted doc is tagged with the schema version at the time of persistence.
194.
by jamwt 15y ago
TBH, we didn't seriously pursue Cassandra when we considered distributed database systems b/c the vast majority of "back-reference checks" we did on the YC network and other area startups was "stay away." We got some very frank advice from
195.
by jamwt 15y ago
> The test hasn't been a "concocted scenario" Btw, b/c I can totally see why you'd read it that way, that particular barb wasn't directed at you, more directed at some of the public benchmarks touted by (non-distributed) NoSQL database
196.
by jamwt 15y ago
All kinds of stuff, mostly ~1-4k values. Read:Write is some low integer, ~2.
197.
by jamwt 15y ago
Cassandra. It's an eventually consistent, fully-distributed database, in the Dynamo mold: http://www.allthingsdistributed.com/files/amazon-dynamo-sosp...
198.
by jamwt 15y ago
Without going too much into specifics and picking on individual databases (which I could do, boy do I have the scars...) When you hit a certain traffic level, scalability, latency and robustness become far more important than single-node op
199.
by jamwt 15y ago
Riak! We're ( http://bu.mp ) using more Riak every day. So far so good.
200.
by jamwt 15y ago
.. or erlang processes. Or python greenlets. Or Haskell forkIO threads. Or libcoroutines.. Etc. See: http://news.ycombinator.com/item?id=1549023
201.
by jamwt 15y ago
Not ignoring your question, btw, but I'm not at liberty to elaborate.
202.
by jamwt 15y ago
Mostly style and idioms. One bug related to inefficiency but not necessarily incorrect behavior. An optimization or two.
203.
by jamwt 15y ago
you're welcome! I did want to make this simple and bulletproof, avoiding dangerous or clever things on the critical path. Particularly wrt the upstream socket. I'm happy you approve. And we definitely have your number in the Rolodex for s
204.
by jamwt 15y ago
.. and I owe much thanks to you for your feedback (and almost all of it was incorporated); I just hadn't asked you for permission to bring your name into the picture, so I didn't! ;-)
205.
by jamwt 15y ago
tl;dr version: https://github.com/bumptech/stud
206.
by jamwt 15y ago
Blocking off 4 /16 networks that were triggering a (self-induced) attack from 10s of thousands of devices sounds pretty distributed to me. (I work at Bump and fought these very Zombies.)
207.
by jamwt 15y ago
Uh, iirc, the assumption here was that the proxy and the application server (and its resources) were not necessarily located on the same physical machine. Serving static content off the proxy is certainly another option, but it involves a
208.
by jamwt 15y ago
Not sure why this is popping up years later, but mrshoe and I presented this at PyCon in 2007. Obviously, the landscape of individual technologies has changed since then, but the basic approaches are still true and pretty language-agnostic
209.
by jamwt 16y ago
They're so different and yet so similar it's probably best taking them one topic at a time. I've written nontrival things in all of those except clojure, but I think I still have a pretty good idea what it's about from a bit of playing aro
210.
by jamwt 16y ago
Well, it's only shared if you share it! I guess it boils down to "don't use mutable globals". That's as true in threaded environments as it is in async ones. If your requirement is that there must be system-enforced safety rather than safe

More ›