b*y
2 楼
不太清楚,不过觉得好像也就是数据存在cloud中(服务器上)。觉得更是个marketing
hype吧。
目前还是延续10多年前的以太网的基础。互联网本身架构没有啥大的变动。
http, tcp/ip, layer 2 ethernet,...
cloud不cloud, 还不是就那么回事儿??
hype吧。
目前还是延续10多年前的以太网的基础。互联网本身架构没有啥大的变动。
http, tcp/ip, layer 2 ethernet,...
cloud不cloud, 还不是就那么回事儿??
b*y
11 楼
SaaS的customerization,
比如某个entity E吧,userA说,我的entityE有属性a1 a2 a3
userB用a1, a3, a4, a5, a6. userC用a2 a4a5a6a8
还要考虑未来user会有其他属性。
那么你怎么model这个呢?
我知道的很多是这样的
定义一个属性table of (attr_name, attr_type, attr_length, isRequired)
然后(user, attr_id)来定义user的属性集
然后就是一个(attr_id, attr_value)的map来存某entity的值
不过这种方法也有不少问题。。。
不知道还有什么其他办法
【在 k***r 的大作中提到】
: SaaS简单吧。application layer的
比如某个entity E吧,userA说,我的entityE有属性a1 a2 a3
userB用a1, a3, a4, a5, a6. userC用a2 a4a5a6a8
还要考虑未来user会有其他属性。
那么你怎么model这个呢?
我知道的很多是这样的
定义一个属性table of (attr_name, attr_type, attr_length, isRequired)
然后(user, attr_id)来定义user的属性集
然后就是一个(attr_id, attr_value)的map来存某entity的值
不过这种方法也有不少问题。。。
不知道还有什么其他办法
【在 k***r 的大作中提到】
: SaaS简单吧。application layer的
k*r
12 楼
这个应该不是cloud computer特有的问题呀。
Lotus Notes时代就有办法了。用比较灵活的data structure,而不是简单的
relational database,或者relational database加上generated schema
或者additional data
【在 b******y 的大作中提到】
: SaaS的customerization,
: 比如某个entity E吧,userA说,我的entityE有属性a1 a2 a3
: userB用a1, a3, a4, a5, a6. userC用a2 a4a5a6a8
: 还要考虑未来user会有其他属性。
: 那么你怎么model这个呢?
: 我知道的很多是这样的
: 定义一个属性table of (attr_name, attr_type, attr_length, isRequired)
: 然后(user, attr_id)来定义user的属性集
: 然后就是一个(attr_id, attr_value)的map来存某entity的值
: 不过这种方法也有不少问题。。。
Lotus Notes时代就有办法了。用比较灵活的data structure,而不是简单的
relational database,或者relational database加上generated schema
或者additional data
【在 b******y 的大作中提到】
: SaaS的customerization,
: 比如某个entity E吧,userA说,我的entityE有属性a1 a2 a3
: userB用a1, a3, a4, a5, a6. userC用a2 a4a5a6a8
: 还要考虑未来user会有其他属性。
: 那么你怎么model这个呢?
: 我知道的很多是这样的
: 定义一个属性table of (attr_name, attr_type, attr_length, isRequired)
: 然后(user, attr_id)来定义user的属性集
: 然后就是一个(attr_id, attr_value)的map来存某entity的值
: 不过这种方法也有不少问题。。。
s*n
17 楼
There's a debate of focus between consistency and availability.
RDBMS as is provided strict consistency but poor availability.
We are entering an era where availability is king.
Sites like amazon and ebay must keep running every single second
of day while consistency can be relaxed with little loss of money.
Can't we have both consistency and availability? CPA theorem
claims that it's impossible; and real life experience with RDBMS
hints that it is very difficult.
Without relational model and tr
【在 b******y 的大作中提到】
: where can I find some in-depth reading?
: clueless to google...
RDBMS as is provided strict consistency but poor availability.
We are entering an era where availability is king.
Sites like amazon and ebay must keep running every single second
of day while consistency can be relaxed with little loss of money.
Can't we have both consistency and availability? CPA theorem
claims that it's impossible; and real life experience with RDBMS
hints that it is very difficult.
Without relational model and tr
【在 b******y 的大作中提到】
: where can I find some in-depth reading?
: clueless to google...
c*t
19 楼
What is "availability" here?
【在 s******n 的大作中提到】
: There's a debate of focus between consistency and availability.
: RDBMS as is provided strict consistency but poor availability.
: We are entering an era where availability is king.
: Sites like amazon and ebay must keep running every single second
: of day while consistency can be relaxed with little loss of money.
: Can't we have both consistency and availability? CPA theorem
: claims that it's impossible; and real life experience with RDBMS
: hints that it is very difficult.
: Without relational model and tr
【在 s******n 的大作中提到】
: There's a debate of focus between consistency and availability.
: RDBMS as is provided strict consistency but poor availability.
: We are entering an era where availability is king.
: Sites like amazon and ebay must keep running every single second
: of day while consistency can be relaxed with little loss of money.
: Can't we have both consistency and availability? CPA theorem
: claims that it's impossible; and real life experience with RDBMS
: hints that it is very difficult.
: Without relational model and tr
k*r
20 楼
also HBase
k*r
21 楼
用过CouchDB吗?偶有些问题。比如说,不是很清楚search怎么做。
全靠real time computation of views?
【在 b*****i 的大作中提到】
: Amazon SimpleDB is not a relational DB, so don't try to consider the
: implementation by that way.
: You can take a look at Apache couchDB which is 'similar' to SimpleDB. And
: search CouchDB and SimpleDB by google.
全靠real time computation of views?
【在 b*****i 的大作中提到】
: Amazon SimpleDB is not a relational DB, so don't try to consider the
: implementation by that way.
: You can take a look at Apache couchDB which is 'similar' to SimpleDB. And
: search CouchDB and SimpleDB by google.
s*n
24 楼
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.
【在 F****n 的大作中提到】
: 这个东西最近有人吵的比较凶,其实我觉得基本上都是废话,因为现代计算机本质上就
: 是线性的,不存在真正的并行,所有的这些保证并行系统consistent的办法,都是以降
: 低效率为前提的,所以搞不懂这怎么就成新大陆了呢
: 这几年很多东西其实就是换了个了WEB SERVICE的马甲。
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.
【在 F****n 的大作中提到】
: 这个东西最近有人吵的比较凶,其实我觉得基本上都是废话,因为现代计算机本质上就
: 是线性的,不存在真正的并行,所有的这些保证并行系统consistent的办法,都是以降
: 低效率为前提的,所以搞不懂这怎么就成新大陆了呢
: 这几年很多东西其实就是换了个了WEB SERVICE的马甲。
F*n
25 楼
现代计算机从理论上说时间和(存储)空间是不可调和的,必须以其中一种来换取另一
种(比如建INDEX),这个所谓的"scalability"实际上最简单的办法就是所有的计算节
点复制一份数据,具体的难点是如何管理这些COPIES,这些东西CS已经研究了几十年了
,没什么新鲜的IDEA。真正比较有点新意的是GRID COMPUTING,可惜有些超前,没有公
司愿意投资。
【在 s******n 的大作中提到】
: 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.
种(比如建INDEX),这个所谓的"scalability"实际上最简单的办法就是所有的计算节
点复制一份数据,具体的难点是如何管理这些COPIES,这些东西CS已经研究了几十年了
,没什么新鲜的IDEA。真正比较有点新意的是GRID COMPUTING,可惜有些超前,没有公
司愿意投资。
【在 s******n 的大作中提到】
: 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.
F*n
27 楼
这种情况应该一边一台机器,两边通过一定的Protocols来建立loosely coupled 的同
步。
Cloud Computing目前来看本质上还是一个Localized Distributed DB,实际上大家讨
厌的EJB就是做这个用的,可惜EJB的这一块被错误的Marketing了。
而Grid Computing才是真正的Virtual Computing系统,从进程管理和资源分配上都跳
出了传统的单机模式,还是比较有意思的。
【在 b******y 的大作中提到】
: 可是application不一定是并行的啊
: 比如AWS可能有一台developer机器,美国时间白天是美国的developer来run
: 晚上就老中老印来run了。相当于租机器挣钱。
步。
Cloud Computing目前来看本质上还是一个Localized Distributed DB,实际上大家讨
厌的EJB就是做这个用的,可惜EJB的这一块被错误的Marketing了。
而Grid Computing才是真正的Virtual Computing系统,从进程管理和资源分配上都跳
出了传统的单机模式,还是比较有意思的。
【在 b******y 的大作中提到】
: 可是application不一定是并行的啊
: 比如AWS可能有一台developer机器,美国时间白天是美国的developer来run
: 晚上就老中老印来run了。相当于租机器挣钱。
b*i
28 楼
Grid computing 很超前吗?感觉都落伍了啊,一些大公司也不提grid了,我所了解的
grid都是基于web service的grid,所以也没太多新意。
【在 F****n 的大作中提到】
: 这种情况应该一边一台机器,两边通过一定的Protocols来建立loosely coupled 的同
: 步。
: Cloud Computing目前来看本质上还是一个Localized Distributed DB,实际上大家讨
: 厌的EJB就是做这个用的,可惜EJB的这一块被错误的Marketing了。
: 而Grid Computing才是真正的Virtual Computing系统,从进程管理和资源分配上都跳
: 出了传统的单机模式,还是比较有意思的。
grid都是基于web service的grid,所以也没太多新意。
【在 F****n 的大作中提到】
: 这种情况应该一边一台机器,两边通过一定的Protocols来建立loosely coupled 的同
: 步。
: Cloud Computing目前来看本质上还是一个Localized Distributed DB,实际上大家讨
: 厌的EJB就是做这个用的,可惜EJB的这一块被错误的Marketing了。
: 而Grid Computing才是真正的Virtual Computing系统,从进程管理和资源分配上都跳
: 出了传统的单机模式,还是比较有意思的。
s*e
29 楼
Good summary. CP sounds really a new bottle with old wine. Let's see how far
it can go with this delimma of consistency and availability.
【在 s******n 的大作中提到】
: There's a debate of focus between consistency and availability.
: RDBMS as is provided strict consistency but poor availability.
: We are entering an era where availability is king.
: Sites like amazon and ebay must keep running every single second
: of day while consistency can be relaxed with little loss of money.
: Can't we have both consistency and availability? CPA theorem
: claims that it's impossible; and real life experience with RDBMS
: hints that it is very difficult.
: Without relational model and tr
相关阅读
[转载] using C++ dll library from java还有人要Gmail吗?大家做JSP/Java开发的配置是怎样的?Interview Questions in Core JavaDuh, Eclipse 3.0 API doc...JSP/SERVLET里面如何实现弹出窗口(新短信)功能JMF Installatin ProblemJAVA 面世题.Re: Can Web Services return Hash or Obje新手上路,请多指教help about bitstream writer从国内连接 mitbbsjavac compilerQuestionwhy doesn't replaceAll work?Java 5.0 is out谁用过JSOBJECT这里有人用Sphinx4吗?GMail giveawayRe: [转载] 急!在线等---从Servlet调用AXIS的Web Service 遇到问题!