BarsMonster

9 ms·

61.
by BarsMonster 17y ago
Ok, then we need DDR3 replicators :-)
62.
by BarsMonster 17y ago
Well, I am not talking 16 channel memory controller :-) 16 slots for 2 memory controller channels. 8 slots per channel. Line count should be reasonable.
63.
by BarsMonster 17y ago
Well, most of low-end 1U servers goes on non-ECC ram, and main problem is still HDD reliability.
64.
Using Atom 330 with server applications(3.14.by)
17 points by BarsMonster 17y ago|12 comments
65.
by BarsMonster 17y ago
I wasn't lost. In fact, original article was initially written in English, and then translated to Russian. It was intentionally corrected according to the taste of the place where it was published in Russian (to reduce "trolling").
66.
by BarsMonster 17y ago
In our extra tests we get 15k real user hits (get an image on a popular web-site front page, 3req/sec average), and then added 2 servers with load generators to get to 50-80 requests per second. We haven't noticed when it auto-scales (that
67.
by BarsMonster 17y ago
The idea of GAE: we do care about scaling instead of you. What I prefer to think: write proper code, and you will never need a cluster (probably not for all cases). There are just 6.5billions people on the earth, so the maximum traffic is l
68.
by BarsMonster 17y ago
I am not denying the fact that it scales. I am saying that on the per-box level it is slow. And if you get slashdotted, you get that "synthetic" load peak, which might not be handled by scaling.
69.
by BarsMonster 17y ago
Yes I've seen it. But it clearly shows some of the problems existing with GAE. It is possible to avoid them, as well as most of the rest, but we cannot deny the fact that there is a problem. Also, there are read-only tests which are also no