x****d
发帖数: 1766 | 1 What are you guys reading or suggest reading/learning for distributed
computing in java? I don't mean big data thing, but JEE.
Solr/Hadoop is using zookeeper. So I guess Zookeeper is a good starting
point. I looked into it a bit, didn't catch much info.
If I have to do something similar, what other options I have other than
using zookeeper to keep my config files in sync? I can think of using JMS/MQ
. Somebody says a look in ZEROmq, but that is something like protocal, to
build something useful, it is long way to go.
So for JEE, if we have to implement something like solr, distributed,
scalable, what would you say is the best bet for next few years? |
w**z
发帖数: 8232 | 2 看不懂，你到底想干啥？
MQ
【在 x****d 的大作中提到】
: What are you guys reading or suggest reading/learning for distributed
: computing in java? I don't mean big data thing, but JEE.
: Solr/Hadoop is using zookeeper. So I guess Zookeeper is a good starting
: point. I looked into it a bit, didn't catch much info.
: If I have to do something similar, what other options I have other than
: using zookeeper to keep my config files in sync? I can think of using JMS/MQ
: . Somebody says a look in ZEROmq, but that is something like protocal, to
: build something useful, it is long way to go.
: So for JEE, if we have to implement something like solr, distributed,
: scalable, what would you say is the best bet for next few years?
|
x****d
发帖数: 1766 | 3 nothing solid at this moment. just want to learn something new.
in the future, I might want to do something keep common data in sync. For
example, now we FTP accounting common data (G/L accounts, cost center, etc)
between ERP/in house systems every day, in the future, I want something like
zookeeper to do synchronized coordination between systems for these common
data. I guess I could use zookeeper but I wouldn't, so I need something to
build something similar.
【在 w**z 的大作中提到】
: 看不懂，你到底想干啥？
:
: MQ
|
g*****g
发帖数: 34805 | 4 All you need is one shared DB for the configuration. We use a service backed
by Cassandra for dynamic properties. |
x****d
发帖数: 1766 | 5 I know, but that is heavy, and old. I want something new, fancy and more
scalable.
For my case, shared DB won't get funding, would get killed as soon as
propose.
backed
【在 g*****g 的大作中提到】
: All you need is one shared DB for the configuration. We use a service backed
: by Cassandra for dynamic properties.
|
g*****g
发帖数: 34805 | 6 Why do you think a shared Cassandra DB is not scalable?
【在 x****d 的大作中提到】
: I know, but that is heavy, and old. I want something new, fancy and more
: scalable.
: For my case, shared DB won't get funding, would get killed as soon as
: propose.
:
: backed
|
w**z
发帖数: 8232 | 7 you need to have an adaptor between erp and in house Db, build service
around the adaptor, maybe some scheduling capability, zoo keeper seems to
be overkill.
There are tons of adaptors for the popular erp.
)
like
common
【在 x****d 的大作中提到】
: nothing solid at this moment. just want to learn something new.
: in the future, I might want to do something keep common data in sync. For
: example, now we FTP accounting common data (G/L accounts, cost center, etc)
: between ERP/in house systems every day, in the future, I want something like
: zookeeper to do synchronized coordination between systems for these common
: data. I guess I could use zookeeper but I wouldn't, so I need something to
: build something similar.
|
x****d
发帖数: 1766 | 8 sorry, 没留意cassandra关键词。cassandra 用zookeeper么？Hbase是用的。
【在 g*****g 的大作中提到】
: Why do you think a shared Cassandra DB is not scalable?
|
x****d
发帖数: 1766 | 9 ZKSS adaptor？
我们G/L 还用mainframe呢，然后乱七八糟无数系统oracle ERP，SAP都有.
我想用一个zookeeper类似的东西做central，写其实是很少的，多数是读，然后给每个
系统做个接口，central变了就跟着变。当然central的源数据还是来自其他各个系统。
to
【在 w**z 的大作中提到】
: you need to have an adaptor between erp and in house Db, build service
: around the adaptor, maybe some scheduling capability, zoo keeper seems to
: be overkill.
: There are tons of adaptors for the popular erp.
:
: )
: like
: common
|
g*****g
发帖数: 34805 | 10 需要啥zookeeper呀，如果在乎consistency，用Quorum读写，或者All写，1读。
如果不在乎几毫秒的consistency，直接1读写就完了。config通常不在乎几毫米的
consistency。
【在 x****d 的大作中提到】
: sorry, 没留意cassandra关键词。cassandra 用zookeeper么？Hbase是用的。
|
|
|
x****d
发帖数: 1766 | 11 好吧，我的use case不需要zookeeper，你给我个use case吧。哪里能用到呢？
你们玩distributed computing都咋玩的？
【在 g*****g 的大作中提到】
: 需要啥zookeeper呀，如果在乎consistency，用Quorum读写，或者All写，1读。
: 如果不在乎几毫秒的consistency，直接1读写就完了。config通常不在乎几毫米的
: consistency。
|
e*****t
发帖数: 1005 | 12 上100个server，你就要zookeeper了，呵呵。
【在 x****d 的大作中提到】
: 好吧，我的use case不需要zookeeper，你给我个use case吧。哪里能用到呢？
: 你们玩distributed computing都咋玩的？
|
x****d
发帖数: 1766 | 13 who said that? be serious, please.
【在 e*****t 的大作中提到】
: 上100个server，你就要zookeeper了，呵呵。
|
x****d
发帖数: 1766 | 14 let us talk about zookeeper, any one using it? |
t*******e
发帖数: 684 | 15 What you need is a data services product. Check out "JBoss Data Services
Platform".
【在 x****d 的大作中提到】
: ZKSS adaptor？
: 我们G/L 还用mainframe呢，然后乱七八糟无数系统oracle ERP，SAP都有.
: 我想用一个zookeeper类似的东西做central，写其实是很少的，多数是读，然后给每个
: 系统做个接口，central变了就跟着变。当然central的源数据还是来自其他各个系统。
:
: to
|
r*****s
发帖数: 985 | 16 这个Jboss data services platform里的
Data virtualization到底virtualize了什么？
看见Jboss metadata repository,
能不能单独用啊？
还有什么好的metadata repository吗？
【在 t*******e 的大作中提到】
: What you need is a data services product. Check out "JBoss Data Services
: Platform".
|
t*******e
发帖数: 684 | 17 多年前jboss上门推销过，是个SOA的产品。基本概念是整合多个data sources，表现成
一个database，virtualization指schema virtulization, 有对应的Hibernate
Dialect。最近没有followup。
【在 r*****s 的大作中提到】
: 这个Jboss data services platform里的
: Data virtualization到底virtualize了什么？
: 看见Jboss metadata repository,
: 能不能单独用啊？
: 还有什么好的metadata repository吗？
|
x****d
发帖数: 1766 | 18 我想的东西类似于这个。大家说说可行不。
http://www.liquid-reality.de/download/attachments/8421391/dosgi_in_details.png?version=3&modificationDate=1360767939000
MQ
【在 x****d 的大作中提到】
: What are you guys reading or suggest reading/learning for distributed
: computing in java? I don't mean big data thing, but JEE.
: Solr/Hadoop is using zookeeper. So I guess Zookeeper is a good starting
: point. I looked into it a bit, didn't catch much info.
: If I have to do something similar, what other options I have other than
: using zookeeper to keep my config files in sync? I can think of using JMS/MQ
: . Somebody says a look in ZEROmq, but that is something like protocal, to
: build something useful, it is long way to go.
: So for JEE, if we have to implement something like solr, distributed,
: scalable, what would you say is the best bet for next few years?
|
x****d
发帖数: 1766 | |
