Use PredictionIO in production

I installed PredictionIO locally, trained the engine using the Universal Recommendation template, which I modified for my needs, everything looks fine.

Now that I know that this can meet my needs, I want to deploy it for production, unfortunately, there is no documentation about it.

Ideally, I would like everything to be deployed on AWS, there is part of the documentation that describes it, but is useless since the CloudFormation template is disabled.

I was thinking about using it, perhaps using Docker to achieve it, but I lack knowledge of the entire stack and would like to understand the following:

  • Where should the data be stored? hbaseIt seems to be a "database", is it dangerous to have it on the same server as the rest (event server, forecast server)?

  • How does it scale? Do I need several instances of PredictionIO working behind a load balancer, or is one enough? if so, how to achieve this?

  • What is a good distributed architecture? to scale, I'm sure we will need to separate the EventServer from the PredictionServer, what is a good way to do this?

Hope someone can help. Thank you Kirill

+4
source share
1 answer
  • Where should the data be stored?

PredictionIO -:

HBase , PredictionIO-0.10.0-incubating/conf/pio-env.sh PIO_STORAGE_SOURCES_HBASE_HOME, HBase.

-

HBase.

  1. ?

predition-io google group,

  1. ?

, , EventServer PredictionServer, ?

, , . . , , , (, ..).

+3

Source: https://habr.com/ru/post/1654796/


All Articles