phoenix-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Stephen Sprague <>
Subject Re: Phoenix newbie questions
Date Fri, 04 Apr 2014 16:28:13 GMT
those are some broad questions but lemme give them a try. And by all means
i'm no expert so take that into consideration too! :)

1. nope. not trigger.  Its just a different API (think sql api) to access
data in HBase.  And all HBase is is an API layer to access data on HDFS. :)

2. according to doc on "create index" not a whole lot.  but do read that
doc.  you'll notice PHX creates another table for the index which is kinda
clever and makes sense the more you think about it.  it also means if you
update your table outside of PHX the index won't be updated. (so don't do

3. well. there's many ways to slice this.  My take is PHX is a (fast) sql
wrapper around HBase which is a "key/val" store used for realtime-y things
(typically.)   Spark/shark (and friends facebook's presto, cloudera's
impala, hortonwork's stinger) are fast wrappers around the genre of "data
warehouse" stores used for analytics (typically.)

these are broad umbrella's of course and use cases cross boundaries all the
time but that's the way i typically categorize them if one must.

make any sense?

On Fri, Apr 4, 2014 at 8:03 AM, Flavio Pompermaier <>wrote:

> Hi to all,
> I'm new to Phoenix and it seems very interesting. I have some questions:
> 1 - from what I understood, Phoenix register itself (like a trigger) to an
> HBase table. Is it right?
> 2 - How much are performance affected by the keeping of a secondary index
> on a mutable table?
> 3 - Which are the difference between Phoenix and Shark/Spark?
> Best,
> Flavio

View raw message