x***n
发帖数: 39 | 1 1. I suppose Oracle store every char(N) as N length,
with padding on, no matter what, so the result you get
back is length N.
2. MySQL and others, sotre every char(N) as N length, (this is obvious),
with padding off when you operate on the content.
e.g. when you INERT INTO c values ('hello'), the data in any RDBMS
is 'hello ... ', which all of us agree.
In operating the content, Oracle pads the data; MySQL and some others
trim the trailing white spaces, so you take back (SELECT *)
1. Or |
|
t**o
发帖数: 620 | 2 【 以下文字转载自 JobHunting 讨论区,原文如下 】
发信人: taro (非典型芋头), 信区: JobHunting
标 题: 急需Java 或者GSM的人才！
发信站: Unknown Space - 未名空间 (Sat Sep 25 18:59:16 2004) WWW-POST
代友发文。
北京一家大型软件外包公司（美资），客户大多是国外
知名公司，比如摩托，西门子等。
急需Java 或者GSM的人才，多个职位空缺。
待遇从优
职位描述：
PC Application Developer (Beijing)
Qualification:
Bachelor degree, Mater degree is preferred.
Major in software or related.
At least 3 years experience in software development.
Excellent in VB, good at C/C++
Good at SQL and RDBMS
Experience in ADO or ODBC
Experience |
|
w*r
发帖数: 2421 | 3 dump 3, 3 make things worse, if you need your data updating can be shown in a
sort of "real-time" manner, 3 is definitely not a choice at all. Logical
Standby would not help your performance issue either, I think your solution
could be seperated into several parts, for the application part, you have to
make optimization to release the connection to RDBMS as soon as possible(as I
described several options in last post) for the server side, to improve the
performance, you could try to optimize the |
|
w*r
发帖数: 2421 | 4 different vendors want their own market share, especially when you seen
database vendors provide their own implementation. the integrity of J2EE and
RDBMS platform just makes it natural to have vendor specific impl, servlet
is not quite so close to the persistence layer. |
|
w*r
发帖数: 2421 | 5 可读性都比较差,有没有人有兴趣开一个blog,或者就在这个版上用wiki的方式做一个好
点的tutorial? 就针对一个RDBMS,例如Oracle, 一个开发平台,例如Eclipse,一个基本
的平台,例如JBOSS,用白手起家方式写一个完整的application, 不用其他的工具或是
framework, eclipse的plugin多且杂，自己一步一步写可以给新手和老手更多地
technical insight, 工具用多了，对于细节的内容通常是不太关注了或者干脆是忘记
了，怎么样？ |
|
t*******e
发帖数: 684 | 6
这个涉及domain specific languages. OO langauges如Java, C#也不是万能的, 比如RDBMS有SQL,
BPM有work flow definition language, spring web flow是个例子, 其他还有jPDL,
OSworkflow等等, ESB有BPEL, rule engine有Rule Markup language, validation engine有
Validation language. 这些大都是XML格式, 但都有一定的logics. |
|
t*******e
发帖数: 684 | 7
也成立。
这个GUI Markup Language就是一种DSL(domain-specific language). 上次和goodbug
讨论过，OO language不是万能的，就如Rule Engine用Rule Markup language, RDBMS用SQL
, GUI design用Declarative programming更直观. |
|
s******n
发帖数: 876 | 8 we do have new type of applications with scale never seen before,
and with consistency requirement very much relaxed. we have each
company with their own gurus and their own ad hoc solutions, but
we don't have a solid theoretical understanding as we do with RDBMS. |
|
c*****t
发帖数: 1879 | 9 templatedb is a relational database, except that that relational schema
is not required up front. It takes on the fact that SQL queries have
absolutely nothing to do with relational schema. Many limitations of
RDBMS as we know it (fixed number of columns etc) are artifacts of
relational schema requirement, which is in turn due to optimization
(which has a lot of assumptions that may not hold).
So templatedb is purely a modification to the data storage system, which
brought the change that elim |
|
m******t
发帖数: 2416 | 10
every
IIUC, the reason OODB never really took off was because of performance
penalty was simply too high.
So it seems to me that we have got the appropriate mix right here -
RDBMS's have got so sophisticate that they are extremely efficient as far
as tabular data processing goes, and then for queries that are truly complex
and dynamic enough to justify an OO-based approach, we have the ORM's. |
|
w*********e
发帖数: 5286 | 11 you must be kidding me, have you every touched any core RDBMS code? |
|
g*****g
发帖数: 34805 | 12 No, but Apache Derby is recognized as a RDBMS everywhere, and
Derby core is only 2M bytecode. It can't be rocket science. |
|
g*****g
发帖数: 34805 | 13 From wiki:
The core of the technology, Derby’s database engine is a full functioned
relational embedded database engine
Not saying Derby is as sophiscated as Oracle, DB2 etc., but Derby demos
the core of an RDBMS regardless. |
|
g*****g
发帖数: 34805 | 14 wiki上写的很清楚是full functional RDBMS. 怕是你的
定义跟别人不一样吧。这个embedded跟embeded OS的embedded
完全不一样，你非扯一块不是显得外行吗。
100 |
|
s***c
发帖数: 1926 | 15 大家都是技术人员，没必要抬杠了。
话说回来，我见过prolog写的数据库内核，很小. prolog虽然慢，但由于CONCURRENY
CONTROL算法与现有RDBMS不同，整体性能相当好。 |
|
w*********e
发帖数: 5286 | 16 只剩下谩骂, 你这种人来谈什么 rdbms kernal真是joke |
|
w*********e
发帖数: 5286 | 17 我也是闲的慌，和你这种Wiki也拿来当标准的的讨论。
还是建议你多看看书把，没事看看mysql， postgrel的code，
看看一个一个rdbms kernal的复杂度到底有多少，
别弄个COURSE PROJ就出来张口判断RMDBS KERNAL的复杂度了。 |
|
w*********e
发帖数: 5286 | 18 对kernal理解不同也挺正常，
但是如果不同意别人对rdbms kernal 的定义，
那说说他自己对kernal的定义好了，
但是什么都说不出，要不就是谩骂，
要不就是拿出wiki上对一个derby的定义来当证据，
有些人只是玩了个玩具汽车（怕是连玩具的原理都没搞明白）
就开始论断汽车的“kernal”复杂度只是一个玩具大小了。 |
|
g*****g
发帖数: 34805 | 19 所谓RDBMS，就是数据存储成表状，对接口的要求一般就是支持
SQL。我举的索引你觉得没用，那就请举一个你觉得公认的定义
大家看看，总不成你的理解就是圣经了吧。
我看你说的这个要吗什么都说不出，要不就是谩骂总结得很好。 |
|
w*********e
发帖数: 5286 | 20 ok, 你的定义（表装数据+sql）能不能称为rdbms kernal我们暂且不论，
就算依据你的定义好了，把什么索引，buffer management, optimizer, statistics，
code generation，全部扔掉, 2M bytecode能支持ANSI SQL的
pipeline execution吗？你对SQL有多少概念？
再一次用你自己的例子，你自己去看看derby，
仅仅就是表装table和sql，derby用了多少code |
|
w*********e
发帖数: 5286 | 21 不是和你说了么，你要学习jiu自己去看看基本的db教科书把，
那上面大部分章节加起来就是一个基本的rdbms kernal.
你自己非要编造自己定义（表装table+sql），别人也没办法.
而且就算你自己的定义，也根本没办法支持你的结论
. |
|
g*****g
发帖数: 34805 | 22 hehe, the entire internet is a waste anyway as there ain't
even an accurate definition for RDBMS core per se. |
|
w*********e
发帖数: 5286 | 23 还什么entire internet，wiki这种可以随便改的上面的一句话也能代表entire
internet了.
说的不好听一点，你对os和rdbms的内核基本上都是一无所知，
我相信你连那derby 那2M bytecode是干什么的也没搞清楚过. |
|
c*****t
发帖数: 1879 | 24 最近没事？跟他争个啥劲？就是一挖坑的。什么是 RDBMS，看看 Codd's
rules 就是了。 |
|
j***i
发帖数: 40 | 25 So you think you can implement all Codd's rules in 2M bytecode(or a course
project, whatever)?
You don't even know the difference between an RDBMS and relational algebra. |
|
F****n
发帖数: 3271 | 26 Does anyone notice that most RDBMSs have unbelievable limits on Date/Time
range they support? For example, mysql only supports date from 1000-01-01,
so you cannot enter earlier dates in a Date/Time column.
What the hell are those providers thinking? Is supporting Date/Time so big
an issue that they have to resort to those stupid limits?
What the solutions most people use? |
|
j*****a
发帖数: 436 | 27 Because MySQL is a free software, that is why they support like that.
Why don't you use other database?
Oracle's date range from 4712BC to 9999AD.
PostgreSQL's date range from 4713BC to 5874897AD. |
|
|
c*****t
发帖数: 1879 | 29 你说的估计一般使用可以足够。但是有些年 +/- 些 second，而如果真需要这种
resolution 的话，还真是个麻烦事。
between
fact
|
|
m******t
发帖数: 2416 | 30
What, are you building some museum inventory management system? |
|
A**o
发帖数: 1550 | 31 and is't only accurate to the second... |
|
F****n
发帖数: 3271 | 32 只要STORE某时刻前后多少毫秒就行了，至于是哪年哪月，应该由一个具体的CALENDAR
来计算。 |
|
g*****g
发帖数: 34805 | 33 这个东西不是有标准的吗？我记得是1970年1月1号？
CALENDAR |
|
c*****t
发帖数: 1879 | 34 那是 unix system 的时间。并不是啥 standard 。 |
|
p*********a
发帖数: 61 | 35 通过 middleware 去控制 transaction 是可能的
但是很复杂
在 rdbms 的设计里，transaction module 和 data storage
是紧密结合在一起的，所以 locking 可以很好的控制 granularity
当把 transaction 独立出来之后，data 对于 transaction 来说就不是透明的
比如对于 range query 来说，究竟要 lock 那些 physical data record,
transaction module 无从而知，只能在逻辑层面来判断
经验上，如果 locking granularity 太大，不能做到 record 的级别，
最终效率会是个非常严重的问题 |
|
z******a
发帖数: 64 | 36 bigtable s3都不算是数据库，只能说是存储。amazon的rdbms是rds，google也有自己
的cloud sql service。如果用ec2又不想花钱用rds，就自己在ec2 instance上装mysql
。用什么数据库根据你的网站规模和需求，你要大规模企业级的一般就是oracle，sql
server，如果就一般的网站mysql就差不多了。 |
|
f*****y
发帖数: 42 | 37 We are looking for a Senior Java Architect who is responsible for
building
the technology used to build the core framework for our various web
initiatives. Ideal candidates will have more than seven years of
professional software development experience designing database-driven,
web-
based systems in a Linux environment.
Office location: Santa Monica, CA
Responsibilities:
• Working with the development team to design robust and scalable
APIs that contribute to the core platform of the ... 阅读全帖 |
|
w****r
发帖数: 945 | 38 JAVA Developer
(Java 6.0, Java EE framework; Spring; Swing, C++; Sybase/DB2; Unix/Linux)
Circa $125,000 + Benefits and bonus package
A very prestigious financial organisation is looking for a Senior Java
Developer to support the design, development and support within their
financial modelling and research area. The Java Financial Model Developer
will be responsible for the development of brand new code, as well as the
enhancement of the existing code base within this market and credit risk
team... 阅读全帖 |
|
f*****y
发帖数: 42 | 39 We are looking for a Senior Java Architect who is responsible for building
the technology used to build the core framework for our various web
initiatives. Ideal candidates will have more than seven years of
professional software development experience designing database-driven, web-
based systems in a Linux environment.
Office location: Santa Monica, CA
Responsibilities:
• Working with the development team to design robust and scalable
APIs that contribute to the core platform of the ... 阅读全帖 |
|
v***e
发帖数: 2108 | 40 在oracle rdbms具体到核心层，SP和你在sqlplus run一个sql是
同一个infrastructure，
具体用多少db的cpu/memory/io,是根据具体你做的事情，不太清楚和
是sp还是不是sp有什么太大关系。
vendor
SP |
|
e*****t
发帖数: 1005 | 41 喜欢这篇的风格。
加一句，也是当年的一个面试题。选择technology的时候，还要考虑整个team的skill
set.
如果大家都是搞spring那一套的，就算ejb更合适，也还是接着搞spring吧。呵呵。
另一方面就是business requirements和budget，对于startup来说，很少会一来就上
jboss,weblogic，websphere.tomcat可以解决的问题就不用上ejb了，反正也不会真把
application跑在不同的node上。
我个人偏好是spring + tomcat，不过ejb3也进化的不错了。
db oracle肯定是enterprise级RDBMS的首选。当然大多数startup是会选mysql的。
ORM不用说，肯定是hibernate了，但也不是所有的application都适合。
NoSQL选择就更多了，具体选择要看requirments。
至于正规军和民兵的analogy，我不太同意。其实都是正规军，哪一个popular open
source framework/tools后面不站着个big name (includin... 阅读全帖 |
|
w*r
发帖数: 2421 | 42 app itself is not a high concurrency OLTP app. the scalability is not a
major concern. Security requires user identify himself/herself while perform
RDBMS operation.
Backend database is currently already configured using LDAP . therefore some
trick needs to be played to achieve higher user experience of not having to
input username./pwd multiple times.
ll
app
e, |
|
w**z
发帖数: 8232 | 43 rdbms, no SQL, different use case. no SQL 牺牲部分ACID, for HA， big
throughput, low latency . CAP theorem |
|
g*****g
发帖数: 34805 | 44 不同系统的难点不一样，相比之下ebay，amazon并发数和可靠性要求都比较高，解决的
方案唯有sharding. Twitter并发极高，需要接近实时的搜索，也很困难，用的是
Cassandra+Lucene这类的解决方案。NoSQL不是为了酷，而是单机数据库搞不定时候的
一种选择，开发要比RDBMS复杂得多。
相比之下，大部分企业应用，对可靠性要求高，但
对并发要求低，实现上反而简单。
IBM和Oracle做得就是这类市场，现有的大量中间件
可以使consulting projects迅速完成。
IBM，Oracle跟这些顶级的web公司比，工资低一些，但稳定，都是有原因的。 |
|
g*****g
发帖数: 34805 | 45 你不说清楚有多大的数据量，多高的并发，read/write intensive, 对transaction和
availability的要求等等，没人能给合适的建议。如果你不知道该用什么，基本上用
RDBMS就对了。 |
|
|
w**z
发帖数: 8232 | 47 everything on Cassandra? Financial transaction gotta be in RDBMS. |
|
w**z
发帖数: 8232 | 48 是，mongo 就存json ,容易上手。Cassandra 是column family +row ,一开始比较难理
解。1.2还弄一个cql 3，very confusing and misleading for the new starters. it
is dangerous to think of Cassandra in a rdbms way. |
|
t***a
发帖数: 416 | 49 对，我觉得这也是mongodb的另一个好处，也真可以把它当rdbms使用，无非就是table+
indexes
it |
|
w**z
发帖数: 8232 | 50 It was pretty eye opening how Netflix trades data consistency for HA and low
latency. They use CL.One for good portion of the cassandra operations.
When I really think about it, it does make a lot of sense. We can compensate
data in-consistency with client side contingency plan. In most of the
cases, we don't really need the strong consistency RDBMS system can offer.
It's all about BIG data now.
for |
|
