vertica udtx JAVA
Hi,
we are on vertica 8.1.0 and JDK 1.8.0.
We have a java udx defined which is defined to aggregate non-numeric type data and we have a fair bit of volume in terms of number of records and also size of the records. At the moment the function is working and we have restricted the planned concurrency of this UDTx to only 2. however our requirement is to open-it up to fairly a higher number of users and the JVM memorysize defined at this moment is 6GB. we want to know if we increase the planned concurrency to 30 and we have 30 users trying to use this function at the sametime then do we need to allocate more memory to JVM and is there a way we can calculate the memroy requirement[basically capacity planning for the JVM resource pool].
another issue we are facing with the UDTx is the data we are trying to aggregate is about 200mb however the JVM memory consumption shoots upto 8GB , is there way to bring down the JVM memory consumption while using the UDTx. Request you to suggest standard practices to be followed while building the UDTx.
Thanks,
Gopalakrishnan.S