d*2 发帖数: 2053 | | b*******d 发帖数: 750 | 2 我觉得不错。和他们聊过。
现在big data for enterprise 挺火的。但有个concern就是这类的小公司太多了,一
打。他们是看着不错。 | l*****t 发帖数: 2019 | 3 我觉得不行。觉得会和datameer一起完蛋。
杰夫。榔头巴克说任何BI要加caching layer的都会完蛋,我粉同意。
【在 b*******d 的大作中提到】![](/moin_static193/solenoid/img/up.png) : 我觉得不错。和他们聊过。 : 现在big data for enterprise 挺火的。但有个concern就是这类的小公司太多了,一 : 打。他们是看着不错。
| b*******d 发帖数: 750 | 4
but real time, or very fast big data joins can be done nowadays, not
necessarily using caching. Dremel and its following work proves a lot, no?
【在 l*****t 的大作中提到】![](/moin_static193/solenoid/img/up.png) : 我觉得不行。觉得会和datameer一起完蛋。 : 杰夫。榔头巴克说任何BI要加caching layer的都会完蛋,我粉同意。
| l*****t 发帖数: 2019 | 5 Not there yet. Drill is beta not doesn't run on apache, impala is out of
beta doesn't run on apache, shark too new. But once these mature, no need
for caching or "smart sampling" from BI tools.That's why I said I don't
think any platfora like company would not last.
【在 b*******d 的大作中提到】![](/moin_static193/solenoid/img/up.png) : : but real time, or very fast big data joins can be done nowadays, not : necessarily using caching. Dremel and its following work proves a lot, no?
|
|