convore.json/groups/epio/controlling-latency/messages.json

[{"user_id": 1, "stars": [], "topic_id": 19931, "date_created": 1303006454.4308641, "message": "But this is a flask app with 3 light SQL queries, it shouldn't take 1.1 seconds IMO", "group_id": 2873, "id": 714862}, {"user_id": 1, "stars": [], "topic_id": 19931, "date_created": 1303006470.1617451, "message": "BTW I'm falling in love with ep.io", "group_id": 2873, "id": 714867}, {"user_id": 1, "stars": [], "topic_id": 19931, "date_created": 1303006681.367074, "message": "So is there anything I can do to help control that latency?", "group_id": 2873, "id": 714893}, {"user_id": 1, "stars": [], "topic_id": 19931, "date_created": 1303006380.4517529, "message": "$ time curl -I http://onefalse.com/results/ericflo\nHTTP/1.1 200 OK\nServer: nginx/0.7.67\nDate: Sun, 17 Apr 2011 02:12:28 GMT\nContent-Type: text/html; charset=utf-8\nConnection: close\nSet-Cookie: session=\"0Buu8Nw4Vl/6zN5heO5ZqpExlPQ=?\"; Path=/; HttpOnly\nContent-Length: 2831\n\n\nreal\t0m1.120s\nuser\t0m0.004s\nsys\t0m0.005s", "group_id": 2873, "id": 714848}, {"user_id": 1, "stars": [], "topic_id": 19931, "date_created": 1303006392.1704209, "message": "$ time curl -I http://127.0.0.1:5000/results/ericflo\nHTTP/1.0 200 OK\nContent-Type: text/html; charset=utf-8\nSet-Cookie: session=\"0Buu8Nw4Vl/6zN5heO5ZqpExlPQ=?\"; Path=/; HttpOnly\nContent-Length: 2424\nServer: Werkzeug/0.6.2 Python/2.6.1\nDate: Sun, 17 Apr 2011 02:12:46 GMT\n\n\nreal\t0m0.025s\nuser\t0m0.004s\nsys\t0m0.006s", "group_id": 2873, "id": 714851}, {"user_id": 1, "stars": [], "topic_id": 19931, "date_created": 1303006417.7636271, "message": "So I totally understand that local should be faster than the one on ep.io", "group_id": 2873, "id": 714857}, {"user_id": 8740, "stars": [], "topic_id": 19931, "date_created": 1303031883.3830819, "message": "@ericflo So, there's a few causes; firstly, apps that are accessed less are swapped out more (just by trying three times I got the latency to drop to half a second)", "group_id": 2873, "id": 716103}, {"user_id": 8740, "stars": [], "topic_id": 19931, "date_created": 1303031940.9959159, "message": "Secondly, our current loadbalancer setup + EC2 networking introduces a base latency of around 0.25 seconds:", "group_id": 2873, "id": 716104}, {"user_id": 8740, "stars": [], "topic_id": 19931, "date_created": 1303031942.2969899, "message": "time curl -I http://www.ep.io/static/favicon.ico\nHTTP/1.1 200 OK\nServer: nginx/0.7.67\nDate: Sun, 17 Apr 2011 09:18:43 GMT\nContent-Type: text/html; charset=utf-8\nConnection: close\nVary: Accept-Encoding\n\n\nreal    0m0.272s\nuser    0m0.000s\nsys     0m0.000s\n", "group_id": 2873, "id": 716105}, {"user_id": 8740, "stars": [], "topic_id": 19931, "date_created": 1303032665.7997301, "message": "but it isn't highly tuned yet", "group_id": 2873, "id": 716163}, {"user_id": 8740, "stars": [], "topic_id": 19931, "date_created": 1303032028.723978, "message": "Obviously we're working to reduce these, in particular by speeding up the DB server, which is the cause of most of the rest of the 0.25 seconds", "group_id": 2873, "id": 716115}, {"user_id": 7, "stars": [{"date_created": 1303076765.831661, "user_id": 1}, {"date_created": 1303078362.7761171, "user_id": 9896}], "topic_id": 19931, "date_created": 1303076513.3783131, "message": "@andrewgodwin you know where to find me if you need MAKE PYTHON MOAR FASTER", "group_id": 2873, "id": 718702}, {"user_id": 8740, "stars": [{"date_created": 1303101692.3938551, "user_id": 8391}], "topic_id": 19931, "date_created": 1303082086.992362, "message": "This probably means I won't get any rest for a long time, but hey.", "group_id": 2873, "id": 719550}, {"user_id": 8740, "stars": [{"date_created": 1303101648.212254, "user_id": 8391}], "topic_id": 19931, "date_created": 1303081627.9256511, "message": "@alex Indeed. My future plans do involve obscenely fast Python. I've heard it's webscale.", "group_id": 2873, "id": 719495}, {"user_id": 1, "stars": [], "topic_id": 19931, "date_created": 1303081668.9224589, "message": "This all makes sense, thanks for the response @andrewgodwin", "group_id": 2873, "id": 719502}, {"user_id": 8740, "stars": [{"date_created": 1303082079.253943, "user_id": 1}, {"date_created": 1303101656.323386, "user_id": 8391}], "topic_id": 19931, "date_created": 1303082071.4136939, "message": "@ericflo No problem. I won't rest till the latency is so low that even Google are envious.", "group_id": 2873, "id": 719547}, {"user_id": 29682, "stars": [], "topic_id": 19931, "date_created": 1303115883.032546, "message": "@andrewgodwin what does the latency profile look like? do you have a particular operation which dominates?", "group_id": 2873, "id": 724185}, {"user_id": 8740, "stars": [], "topic_id": 19931, "date_created": 1303123122.7537661, "message": "@bradwhittington As I said above, it's a roughly even split between 250ms in the loadbalancer and a slightly slow database server", "group_id": 2873, "id": 724852}, {"user_id": 7, "stars": [], "topic_id": 19931, "date_created": 1303136560.8247881, "message": "@andrewgodwin we should get you to the goethenberg sprint to work on eventlet on pypy :)", "group_id": 2873, "id": 726596}, {"user_id": 8740, "stars": [], "topic_id": 19931, "date_created": 1303151299.8269269, "message": "@alex I would love to get greenlets working on pypy, haven't even tried writing rpython n a while though", "group_id": 2873, "id": 729055}, {"user_id": 1081, "stars": [], "topic_id": 19931, "date_created": 1303275546.2156129, "message": "I'm obviously late too this, but its my one complaint, i have very little used apps, low traffic, for a client/friend, but he kept noticing that it takes a bit of a while to load the first time", "group_id": 2873, "id": 752111}, {"user_id": 8740, "stars": [{"date_created": 1303320891.6442649, "user_id": 1}, {"date_created": 1303339867.081531, "user_id": 1081}], "topic_id": 19931, "date_created": 1303296918.838979, "message": "@rasiel That's one of the trade-offs of our approach; we're going to offer the ability to keep the app even more warm and ready to go, but that will have to cost, as it'll take up RAM all the time.", "group_id": 2873, "id": 753647}]