k*****a
发帖数: 1463 | 1 Go back and learn why HADOOP (BASE: basically available, soft state,
eventual consistency)?
：所以 到了这个时候，就只有两种选择。一是NoSQL DB，二是Sharding。
一是NoSQL DB ？？？
>> 瞎腚: EBay's items don't have much consistency contentions, would be able
to be handled by some garage DBMS after proper 瞎腚, 别说SQLServer.Ebay's
transaction isn't at all scale a problem. How many
bid items live everyday? Are you suggesting Ebay to use NoSQL DB for live
site transactions？and you put these as your first option to scale out EBAY?
Read... 阅读全帖 |
|
g****n
发帖数: 3370 | 2 这个问题已经让我纠结了两个月了，实在无计可施了。请教一下各位看看有什么建议没
有。
先谢过了！
我用的DSL上网，dry-loop, modem工作在bridge状态，router负责PPPoE及其他。
用了几年好好的，但从两个月前开始，出现断线情况。症状是，突然没有征兆
的PPPoE断开，但modem的DSL灯还亮着，看它的状态依然是sync的，看log也看不出
有什么不正常。但看router的log，断线时出现这样的信息：
LCP appears to be disconnected
Sending PADI
Resending...
All WAN down. Reconnecting ...
也就是说router发出PADI但接收不到PADO。这种情况多发生在正在BT下载时，比如
今天，已经连线2天多了，我从5点开始下载，结果在10点发生状况。
解决的办法只有一条：重启modem。重启后马上连上。
已经换掉router和modem，新的问题依然存在。所以基本可以排除router/modem问题。
关于线路，在发生问题前后线路没动过，而且modem状态显示SNR margin和衰... 阅读全帖 |
|
d******i
发帖数: 7160 | 3 偶把地沟油刷成番茄了。N-Only，40M带宽。upper range。Automatic选channel。
Status里发现自己选的Channel是1.
试着手动设Channel，scan了先。
惊奇发现只暴露出Channel 5-13(选lower range则会暴露出channel 1-9).
只有选20M带宽才能暴露出全部的1-13。
这是为什么呢？
还有，in this case,番茄给自己选了个channel 1，scan结果又没暴露出channel 1的
db数，
那么我靠什么去知道是否5-13里面那个最好的channel(负数dbm值最大的那个)是否比
channel 1要好呢？
谢谢指教！ |
|
i*********e
发帖数: 1010 | 4 MAC OSX
OPTION+点击wifi的Icon
精确度dbm
★ 发自iPhone App: ChineseWeb 7.8 |
|
c*****h
发帖数: 14923 | 5 民用那没办法了，AC+FIT AP太贵
只能尽量设到最好的方案
几个router chanel 选择互不干扰的1 6 11，然后RSSI 低于 -70 dBm就断开
AC68U有无线漫游功能，比其他神油好点吧，虽然不能完全无缝
iphone有这点好处：只要ssid密码都一样，iPhone会自动选择信号好的进行连接
我家基本都是果子产品，所以使用起来毫无问题 |
|
l**********l
发帖数: 209 | 6 不是完美的无缝，会先断开然后再连接，大家2，3秒吧。路上可以设置断开的dbm下限 |
|
e****y
发帖数: 26 | 7 In doing a course project, i need to create table, insert data into a DBMS,
I wonder which interface is better?
JDBC or SQL-C? Which is easier or more powerful?
I have some experiences with Pro*C(C/C++ interface of Oracle) but never
touch JDBC before.
Any suggestion is appreciated. |
|
xt
发帖数: 17532 | 8
if
Oracle JDBC driver, especially the Oracle:thin driver, is the worst I know
amoungst
major DBMS's. It has many many problems. For example, if you define a CHAR[40]
field
in your schema, then set it to "hello". It works fine if you do a direct SQL
search
for WHERE field='hello'. However if you use JDBC with the very same string,
you will
never get the record back. You have to use "hello" padded with 34 spaces to
get the
record back. Talking about consistency! IMHO, Sybase has the best SQL
comp |
|
xt
发帖数: 17532 | 9
Come on! It doesn't happen with some commercial DBMS and you
expect that kind of friendly error message with an open source
freeware? |
|
xt
发帖数: 17532 | 10
I only use commercial DBMS, which is already
free to me. |
|
xt
发帖数: 17532 | 11
ignored?
Sybase, DB2, Oracle all suck in JDBC documentation.
Maybe that problem is already fixed. Try Oracle thin driver. What happens is
that
if you insert a string less than 32 long, the DBMS will automatically append
white
space to that string and then store into DB. There is nothing wrong with this
behaviour. However, when you query that table with the original string you
used,
you will not get anything back, unless you append white space to make that
string
32 long.
OK, there is nothing wr |
|
w*r
发帖数: 2421 | 12 come on, you guys are complainning about oracle, that
darn piece of dbms's JDBC is much better implemented comparing
to most others. I am using TeraData everyday, all kinds of issues
comes from the JDBC part. I was originally thought the company
like NCR whose customers usually pay hundreds of thousands of
dollars each year for just single node support should have a much better
implementation of JDBC, guess what? their own JDBC driver
cannot even parse some the legitimated SQL statement which
ru |
|
w******t
发帖数: 1422 | 13 I see - what you described still exists in 9i - thanks
but this is NOT a bug - in my opinion - and I actually would expect this kind
of behavior if you use fixed length of string def.
if you want variable length, you should go with varchar -- just my opinion
regarding jdbc driver overall - oracle sucks, even though it's much better
than some other dbms driver, it still sucks. Only diff is that other DB driver
sucks "bigger" time. :)
however, we have to live with that - I guess
in
this
the |
|
xt
发帖数: 17532 | 14
Let me explain what CHAR(32) means. CHAR(32) means a fixed length string
field as opposed what is defined by VARCHAR or CLOB etc. However, CHAR(32)
does not mean you have to make your value 32 long, but you cannot go beyond
32. In case you do not have 32 characters, DBMS will append white space
to it - which is what Oracle is doing as well. However, that does not mean
that when you retrieve the data, it should give you the white space it adds.
Obviously MS, IBM, Sybase and even Oracle SQL Plus |
|
T*****e
发帖数: 361 | 15 我主要是想用JSP做Web开发，现在正在看Java的教程，正准备配置编程环境。
能够想到的包括：
1 J2EE/J2SE 1.4.01，这两个选哪一个好呢？JSP需要有Application Server么？
是不是J2SE+别的东东也可以呀？呵呵，没人理我的话，就装J2EE，多就多点吧。
2 Eclipse 3.0，还需要什么plug-in么？
3 Apache 2.0.50，TOMCAT是直接在里面了吧？如果不在，就另外安装吧，
应该能够搞定。准备装在自己的机器上(WinXP Home)，这样可以直接调试了。
不过Network Domain和Server Name两项不知道该怎样设置？
4 DBMS，准备用MySQL。
这样配置主要是为了能够在自己的计算机上做开发和调试。
开发差不多之后，会考虑往别的server上移植，可能会试Linux和WinXP吧。
各位大侠的开发环境是怎样配置的呢？对于我的初步想法，欢迎任何建议和评论。
虽然自己折腾，可能最后也能够鼓捣出来，不过大侠们的任何提示，都可能使我
少走不少弯路。多谢了先。 |
|
xt
发帖数: 17532 | 16
才10,000? 我认为1000的Connection pool足以胜任.随便哪个DBMS
都能开这么多. |
|
xt
发帖数: 17532 | 17
才10,000人,我就不信搞不定.1000的connection pool都浪费.我认为最大256足够用.
我就不信10000人提交数据的速度居然比DBMS存储速度的1/40还快.
我办公室机器1G MEM,开256个connection根本没有问题. |
|
l*****e
发帖数: 1431 | 18 Sorry for being clear at the first place. I am trying to implement my own JDBC
driver for Oracle DBMS. It talks with Oracle CLI (as I understand what a type
4 JDBC driver should do) and provides the standard JDBC interface specified by
Sun to other Java applications.
了， |
|
Q**g
发帖数: 183 | 19 哈，这个方便。
估计实现也就是调一调DBMS specific的SQL了，呵呵 |
|
F****n
发帖数: 3271 | 20 I think it can only be done with some DBMS extension. However, before you do
that, you should realize this is NOT new stuffs. OODBMS has handled that a
long time ago. It's a shame that fat cats in big companies turned down
OODBMS, and that's the reason today we need those stupid ORM stuffs. |
|
m******t
发帖数: 2416 | 21
The idea behind pure java dbs is (obviously) not to compete with
those more mature C implementations as a generic dbms solution.
Pure java-based dbs are mostly designed to be embedded in applications,
or to serve as testing fixture. |
|
w*********e
发帖数: 5286 | 22 我觉得严格说，dbms不是仅仅是os之上的application，
如果真的要上规模的话，要改很多os里面的东西。
用java写可能会有很多限制，当然，如果只是
小规模的应用的话，用java也可能满足要求了。 |
|
m******t
发帖数: 2416 | 23
Finally... I didn't want to say anything because I normally don't
pick on people's spelling, but I really think that if people choose
to post in a condescending tone and keep referring others to
"basic dbms textbooks," they should probably also make sure
they got right the one keyword that the whole debate is about. |
|
F****n
发帖数: 3271 | 24 The step I took: directly write SQL to create tables in the DBMS. You can
create all the indexes there while JPA cannot.
I mean, it's not the point. |
|
s***8
发帖数: 1136 | 25 When creating a table with one column type as VARBINARY, got the following
error:
ERROR 42X01: Syntax error: Encountered "" at line 1, column 76.
Issue the 'help' command for general information on IJ command syntax.
Any unrecognized commands are treated as potential SQL commands and executed
directly.
Consult your DBMS server reference documentation for details of the SQL
syntax supported by your server.
IF I change varbinary to varchar, it works. But I need to use varbinary.
Anyone can help? |
|
F****n
发帖数: 3271 | 26 If you think of DBMS as nothing but indexing, Lucene has its own indexing
managing & access mechanism, which is much faster than other DBs for Lucene'
s own specific tasks. |
|
S*******h
发帖数: 7021 | 27 【 以下文字转载自 JobHunting 讨论区 】
发信人: StanMarsh (Stan), 信区: JobHunting
标 题: 多個Software Engineer 職缺
发信站: BBS 未名空间站 (Mon Sep 10 20:35:35 2012, 美东)
Multiple software developer and software engineer in test openings. Send me
a message if you are interested in.
Software Engineer Description
Would you like to use your expertise to help create exciting new cloud-based
products and services? The successful candidate will be working with a
small engineering team to design and build high performance, scalable ... 阅读全帖 |
|
F****n
发帖数: 3271 | 28 大部分实际应用都不是标准SQL,
各个DBMS的naming rule, quotation, data type都不一样
很多公司都会经历DB Migration (e.g., mysql -> Oracle),
JPA/hibernate很轻松，JDBC/SQL pain in the ass |
|
c*****t
发帖数: 1879 | 29 你还是迷糊了。这道题不是考 partition 。这个是 trivial 。否则就直接说
offline 就可以了。
Availability 在 database 这一行指很多东西。但是在 database 里面
data integrity 是最重要的。
那么综合起来，就有一些比较 tricky 的地方。比如
rowA 刚刚从 machine1 弄到 machine2 。
这时候，database 收到 update / delete rowA 的 request 。这个怎么办？
同样的，machine3 刚刚收到 rowA，还没来得及告诉 machine1 ，这时候
DBMS 收到关于 rowA 的 update / delete request 。
不错，简单的 locking 似乎可以工作。但是，如果 machine1 / machine 3
这时候刚好 crash / restart ，那么如何保证 data integrity ？如何不让
整个系统因为一台机器 crash （比如 machine 3），而 hang ？
这道题如果没有考虑这些问题，基本上就 fail 了。 |
|
x****u
发帖数: 44466 | 30 所以我说嘛，要理解文科mm们痛恨DBMS的心情。 |
|
n*****t
发帖数: 22014 | 31 所谓看到能买到这个定义有歧义：查询的时候有票买不到，还是票源锁定后买不到？
前者当然没法保证也没人能保证，这个是 biz logic 决定的不是系统做不到（查询的
时候就给你锁票？）
后者除非 server 当机了，否则不会有问题。如果是前台当掉了，重新刷一次就没事了
，这个估计谁都习惯了，哈哈。
现在实际的问题，就是都在一个节点查询、锁票，一个大的 DBMS 显然搞不定，应该把
bottleneck 拆出来。 |
|
n*****t
发帖数: 22014 | 32 哎，真不知道说你啥好了，16k 是我的烂笔记本，离极限性能差老远老远，而且绝对比
现成 DBMS 强几条街。抛开这些不说，我北京局的 DB B 放一个单独 server 行不行？
再不够一个车次一台 server 行不行？无非前段计算一下应该向谁申请。我这可是实时
的哦，假设算力足够。
你们啊，只会用别人做好的工具，所以才会把一个破玩意捧上天
16k |
|
r***y
发帖数: 4379 | 33 多谢朋友, 一语点醒.
按照 goodbug 的提示, 能否容易地 scale out 就不是在 MySQL 和 PostgreSQL 中作
选择的条件了. 做 scale out 的话, MySQL 也好, PostgreSQL 也罢, 两个都是 "
Relational DBMS" , 如果你不熟悉的话, 都不容易. 如果你熟悉其中一个, 肯定弄起
来会比另外一个要容易.
这样看来, 要比较也是比较其它 features 了.
不过菜菜鸟的观点, 既然用开源, oracle owned 就该划掉了, 再考虑到 mariaDB 的满
身铜臭... 也就没有原因不选 PostgreSQL 了. |
|
