Re: How to speed up write performance

2017-09-08 Thread James Taylor
Here's another good tuning resource that covers HBase too: http://phoenix.apache.org/presentations/TuningForOLTP.pdf On Fri, Sep 8, 2017 at 8:30 AM, Josh Elser wrote: > Hef -- do your split points actually correspond with the distribution on > values of your `id` column? You

Re: How to speed up write performance

2017-09-08 Thread Josh Elser
Hef -- do your split points actually correspond with the distribution on values of your `id` column? You can tell this pretty easily looking at the number of requests per region for your data table on the HBase UI. And yes, PQS will not increase the performance (as it is adding "more work" to

Re: How to speed up write performance

2017-09-08 Thread Hef
Hi James, I have read over the Tuning Guide, and tried some of your suggestions: #3, #5, #6. Since the date is mutable, and read/write frequently, I did not try #1, #2, #4. The schema is simple as such: *create table if not exists test_data (* * id VARCHAR(32),* * sid VARCHAR(32),* * uid

Re: How to speed up write performance

2017-09-06 Thread James Taylor
Hi Hef, Have you had a chance to read our Tuning Guide [1] yet? There's a lot of good, general guidance there. There are some optimizations for write performance that depend on how you expect/allow your data and schema to change: 1) Is your data write-once? Make sure to declare your table with the

How to speed up write performance

2017-09-05 Thread Hef
Hi guys, I'm evaluating using Phoenix to replace MySQL for better scalability. The version I'm evaluating is 4.11-HBase-1.2, with some dependencies modified to match CDH5.9 which we are using. The problem I'm having is the write performance to Phoenix from JDBC is too poor, only 500writes/second,