p***c 发帖数: 5202 | 1 哪个运行维护比较easy些?Thanks
我看了下,仿佛MongoDB容易些 |
a*w 发帖数: 4495 | 2 在这两个里面,我感觉MongoDB文档好一点。
【在 p***c 的大作中提到】 : 哪个运行维护比较easy些?Thanks : 我看了下,仿佛MongoDB容易些
|
z*******3 发帖数: 13709 | 3 mongodb的话据说要把整个index都读入内存?
还是cassandra吧 |
z*******3 发帖数: 13709 | 4 目测了一下,nosql主要选择是cassandra和hbase
前者是facebook的东东,后者是waterloo和hadoop的东东
mongodb貌似已经落伍
网络上可以找到大量的nosql war cassandra vs hbase的文章
hadoop我看了下,还有大幅提升的空间
现在的机制还是太过于复杂,spring可以对hadoop做一定的优化
这些东西绝大多数都还在0.x和1.x版
普遍不太成熟,还有不少路要走 |
w**z 发帖数: 8232 | 5 Cassandra or hbase. Cassandra 好用些,但favor writes . Hbase is more
balanced. mongo 太占内存。
【在 p***c 的大作中提到】 : 哪个运行维护比较easy些?Thanks : 我看了下,仿佛MongoDB容易些
|
k********e 发帖数: 368 | 6 Cassandra 爱好者前来围观。 无限可扩容。 |
p***c 发帖数: 5202 | 7 Thanks
看来都不喜欢mango啊,呵呵。我主要是看了点教程,发现mongodb仿佛很简单,几下就
配置好,可以写程序了。
Cassandra我到datastax下载了他们的enterprise server,安好,跑了个demo,发现和
mongo很多不一样 |
t*******e 发帖数: 684 | 8 这3个之中,哪个可以做OLTP,哪个适合做OLAP? |
t***a 发帖数: 416 | 9 我喜欢mongo, 够简单,数据拿出来直接是json
和ror整合的也好,scala/java的driver也不错
缺点大家也说了,就是这厮对内存喜欢多吃多占,而且如果它内存如果抢不过别人,
performance就急剧下降,但这策略也挺和我胃口的。。。现在我们mongodb的server就
只装mongdb,内存就随它搞吧。。。。
【在 p***c 的大作中提到】 : Thanks : 看来都不喜欢mango啊,呵呵。我主要是看了点教程,发现mongodb仿佛很简单,几下就 : 配置好,可以写程序了。 : Cassandra我到datastax下载了他们的enterprise server,安好,跑了个demo,发现和 : mongo很多不一样
|
w**z 发帖数: 8232 | 10 别的不知道,Cassandra has intergration with Hadoop
【在 t*******e 的大作中提到】 : 这3个之中,哪个可以做OLTP,哪个适合做OLAP?
|
|
|
w**z 发帖数: 8232 | 11 自己写写玩和在production ,几十数百node 一起跑,两码事。
【在 p***c 的大作中提到】 : Thanks : 看来都不喜欢mango啊,呵呵。我主要是看了点教程,发现mongodb仿佛很简单,几下就 : 配置好,可以写程序了。 : Cassandra我到datastax下载了他们的enterprise server,安好,跑了个demo,发现和 : mongo很多不一样
|
w**z 发帖数: 8232 | 12 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.
【在 t***a 的大作中提到】 : 我喜欢mongo, 够简单,数据拿出来直接是json : 和ror整合的也好,scala/java的driver也不错 : 缺点大家也说了,就是这厮对内存喜欢多吃多占,而且如果它内存如果抢不过别人, : performance就急剧下降,但这策略也挺和我胃口的。。。现在我们mongodb的server就 : 只装mongdb,内存就随它搞吧。。。。
|
t***a 发帖数: 416 | 13 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
【在 w**z 的大作中提到】 : 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.
|
w**z 发帖数: 8232 | 14 Mongo comes with Morphia, 可以做类似sql 的操作,不知性能如何? Cassandra 是
存成sstable, 有些操作不可能实现。
table+
【在 t***a 的大作中提到】 : 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
|
t*******e 发帖数: 684 | 15 哪个能干real time, ad hoc query?
【在 w**z 的大作中提到】 : Mongo comes with Morphia, 可以做类似sql 的操作,不知性能如何? Cassandra 是 : 存成sstable, 有些操作不可能实现。 : : table+
|
g*****g 发帖数: 34805 | 16 Cassandra is fast, but you need to plan your query. |
w**z 发帖数: 8232 | 17 depends on what you want to do. think of Cassandra as a big hash table,
value is a list of columns (name value pair again ) ordered by column name.
【在 t*******e 的大作中提到】 : 哪个能干real time, ad hoc query?
|
p*****2 发帖数: 21240 | |
p***c 发帖数: 5202 | 19 columnfamily是row的意思?
column感觉就是相当于excel里面一个cell?
it
【在 w**z 的大作中提到】 : 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.
|
w**z 发帖数: 8232 | 20 CF is collection of rows , row is collection of columns. treat it as
hashtable inside of a hashtable .
【在 p***c 的大作中提到】 : columnfamily是row的意思? : column感觉就是相当于excel里面一个cell? : : it
|
|
|
w**z 发帖数: 8232 | 21 不懂。
【在 p*****2 的大作中提到】 : CouchDB是啥情况呀?
|
t*******e 发帖数: 684 | 22 Think about a scenario at amazon.com, when a pricing error occurs, the
amount of txns jacks up with anomaly against a single item. Obviously,
overnight data analysis/mining doesn't help. There must be a real-time, big
data analytic process to quickly remove the item from listing.
Does Cassandra or HBase fit this use case?
.
【在 w**z 的大作中提到】 : depends on what you want to do. think of Cassandra as a big hash table, : value is a list of columns (name value pair again ) ordered by column name.
|
F****n 发帖数: 3271 | 23 Because the DB is too good, Amazon received tons of orders at the wrong
prices.
On the other hand, my crappy server could have refused those connections
with no problem:))
big
【在 t*******e 的大作中提到】 : Think about a scenario at amazon.com, when a pricing error occurs, the : amount of txns jacks up with anomaly against a single item. Obviously, : overnight data analysis/mining doesn't help. There must be a real-time, big : data analytic process to quickly remove the item from listing. : Does Cassandra or HBase fit this use case? : : .
|
e*****t 发帖数: 1005 | 24 not very familiar with hbase, but cassandra would be fine in this scenario.
Cassandra have very good write performance, for read, the performance depend
s on the consistency level you need.
In your use case, the consistency doesn't matter much.
big
【在 t*******e 的大作中提到】 : Think about a scenario at amazon.com, when a pricing error occurs, the : amount of txns jacks up with anomaly against a single item. Obviously, : overnight data analysis/mining doesn't help. There must be a real-time, big : data analytic process to quickly remove the item from listing. : Does Cassandra or HBase fit this use case? : : .
|
g*****g 发帖数: 34805 | 25 This fits to monitoring instead. And many existing tools can tell you a
particular url/service is hot.
big
【在 t*******e 的大作中提到】 : Think about a scenario at amazon.com, when a pricing error occurs, the : amount of txns jacks up with anomaly against a single item. Obviously, : overnight data analysis/mining doesn't help. There must be a real-time, big : data analytic process to quickly remove the item from listing. : Does Cassandra or HBase fit this use case? : : .
|
t*******e 发帖数: 684 | 26 Machine learning/data mining tools can do the job. Screening spam is another
use case, but much simpler. The challenge lies in the aggregate function.
To make ad hoc queries relying on an aggregate function that touches the
entire datastore real-time is really tough, if not impractical.
【在 g*****g 的大作中提到】 : This fits to monitoring instead. And many existing tools can tell you a : particular url/service is hot. : : big
|