vicaya

24 ms·

181.
by vicaya 17y ago
I think I independently came up with the same idea to solve the distributed gfs master problem: use a separate bigtable/gfs1 cluster for master metadata for gfs2. I'm glad that I'm not crazy :)
182.
by vicaya 17y ago
Is this a mylivesignature.com knock off?
183.
by vicaya 17y ago
Core developer of Hypertable ( http://hypertable.org/ ). Had patches accepted in Git, Reviewboard, Thrift, etc.
184.
by vicaya 17y ago
I think Twitter's model (following and followers) is appropriate for HN. The traditional friend model (friendster, myspace, facebook etc.) is ugh.
185.
by vicaya 17y ago
Once singularity is reached, our code will write themselves. We'll just sit back and enjoy a latte/beer/whatever.
186.
by vicaya 17y ago
google for 1and1, you'll see a lot of complaints there. I had a few bad experience with them. If your renew payment has some issue (mostly CC expiration date etc.) they'll immediately send the account to collection and suddenly your $6.99 o
187.
by vicaya 17y ago
Cloud computing provides service abstraction at different levels. You don't need to think about servers at SaaS (Software as a Service. e.g, Google Docs/Apps) and PaaS (Platform as a Service. e.g. Google Apple Engine, Windows Azure; Amazon
188.
by vicaya 17y ago
Call me old school, but I don't like having desktop environment in a browser (webtop sounds fine to me) to be called an OS. An OS is much more than that.
189.
by vicaya 17y ago
Erlang is a quirky DSL for a seasoned actor system. That's my impression 10 years ago. It still makes sense now despite all the recent hypes.
190.
by vicaya 17y ago
I still think SSD is a mere diversion in storage. According to Jason Hoffman (CTO/Founder of Joyent, speaking on Structure'09), SSD under their typical workload can only last for a month before wearing out: http://gigaom.com/2009/06/25/str
191.
by vicaya 17y ago
99.9% web sites are CRUD, which don't really need full features of RDBMS, which is both an overkill and a bottleneck for scalability. NoSQL is ironic word in that both Hive and Pig are gonna introduce an ANSI SQL dialect for scalable data q
192.
by vicaya 17y ago
It's poor man's Vertica. Mostly good for analytics workloads. It's quite strange that they didn't reference Bigtable paper at all, while saying "to the best of our knowledge, there exists no published deployment of a parallel database with
193.
by vicaya 17y ago
Actually, you can write a wrapper in your favorite language around your favorite database and call it your favorite filesystem, which is really a CMS.
194.
by vicaya 17y ago
Well, one of the best research in compilers and languages, IMHO is applying static analysis and constraint solvers to practical systems (e.g. C) find bugs and/or automatically generate tests to find bugs: http://www.stanford.edu/~engler/
195.
by vicaya 17y ago
Dude, these guys love analogies like: Astrology is to fortune as IQ is to achievement.
196.
by vicaya 17y ago
The "symmetrical" is the killer part. It makes peer to peer backup of large files (HD videos) bearable.
197.
by vicaya 17y ago
Wow, astounding ignorance/ungratefulness on static analysis. Coverity (a static analysis tool) found many defects in the code (17 fixed). The number of defects per KLOC for SQLite is not that great compared with other projects. KDE is actua
198.
by vicaya 17y ago
search for "dvcs comparison" (unquoted). bing returns many completely unrelated dvd/spam sites, while google is spot on. MS still have a lot of work to do.
199.
by vicaya 17y ago
It's entirely possible to implement a wave like UI over SMTP. Many mail clients already implement thread views etc. If you take a step further and implement a threaded diff view it's already fairly close to wave UI). While I like the the sh
200.
by vicaya 17y ago
Read the facebook engineering notes on memcached. Linux kernel needs to be patched to fully utilize all cores even at 200k qps with 173ms average latency. Also you didn't give average request size and latency numbers, which makes your numbe
201.
by vicaya 17y ago
A typical "drag race" benchmark. There is no latency numbers for the corresponding throughput. The facebook engineering notes was a little more helpful, with at least average latency 173ms at 200k qps with an intel 8-core box with 1GE NIC (
202.
by vicaya 17y ago
Ubiquitous browser support so that the sign on interface is uniform and that phishing is impossible would go a long way for OpenID.
203.
by vicaya 17y ago
Sorry, but 500MB DB size is a tiny dataset these days (anything < 1GB is tiny, < 4GB is small, < RAM on a single node (~8GB-64GB) is medium, < Disks on a single node (~128GB to a few TB) is large, huge dataset requires multi
204.
by vicaya 17y ago
"Done" is overrated. It just gives you false/fake sense of control, which can be addictive. Give it up. "Doing" instead.
205.
by vicaya 17y ago
Bigtable (according to the paper), HBase and Hypertable all use bloom filters to improve performance of random reads. Here is some bloom filter test code in Hypertable to compare Jenkins' lookup3, Hsieh's SuperFastHash and Appleby's MurmurH
206.
by vicaya 17y ago
Memcached with UDP get can handle 400k gets per second (from a facebook engineering note earlier this year) on a single commodity server. Which "decent" DB can even approach this even for simple queries? A cache server and DB server simply
207.
by vicaya 17y ago
Why stop at average karma, we can also print a little histograms (sparklines) of comment karma, I bet the target audience of HN would love it :) Another way for front-page ordering would be (total karma * exp(-beta * age)), you can tweak be
208.
by vicaya 17y ago
What you need is a full text search engine with faceting features. Solr or Sphinx should fit the bill.
209.
by vicaya 17y ago
http://www.wolframalpha.com/WolframAlphaAPI.pdf works and more up to date (updated on May 14, 2009) Looks pretty good.
210.
by vicaya 17y ago
He's the main developer of stackoverflow, which I think is pretty good compared with many other sites :)

More ›