Samarth, Gunnar, our Chief Technologist on DSM can answer all your questions but briefly: 1. The Phoenix log table can be a source for the central repository but the central repository is in Vertica (you can use the community edition). This could be any database and we are working on making Trafodion the repository soon. It requires making all the DDL and DML work with this new Central Management Instance (CMI) repository. 2. Various means can be used to access the Data Service Instance (Phoenix) data to map to the normalized schema of the CMI. When Gunnar presents it to you he can cover these mechanisms. 3. The UI dashboards and whiteboard provide these capabilities against the CMI. Vertica's analytical support helps with that. Trafodion has most of that support but would require changing some queues and pulling some of it into the client. 4. We would like to open source it but have not made that decision as yet. There are potential open source models short of open sourcing all of it. But the pricing model for it in the interim is an open source pricing model.
Rohit On Feb 11, 2015, at 8:49 PM, Samarth Jain <[email protected]<mailto:[email protected]>> wrote: Hi Rohit, I am interested in knowing more about DSM and its capabilities. I have some questions: 1) Phoenix logs the traces in a Phoenix table. Can a Phoenix table work as a DSM repository? 2) Does DSM expect data to be in a particular format or does it provide a means to plug in you own data puller (JDBC, REST etc)? 3) Does it provide an analytical engine that can roll up data (among other operations) and slice and dice it on various dimensions? 4) Is it open sourced? From the email it sounded like it isn't. Thanks, Samarth On Wed, Feb 11, 2015 at 11:11 AM, Tapper, Gunnar <[email protected]<mailto:[email protected]>> wrote: JMX interfaces are fine, too. Thank you, Gunnar Download a free version of HPDSM, a unified big-data administration tool for Vertica and Hadoop at: HP DSM Download<https://vertica.hpwsportal.com/#/Category/%7B%22categoryId%22%3A10185%7D/Show> “People don’t know what they want until you show it to them… Our task is to read things that are not yet on the page.” — Steve Jobs From: Jain, Rohit (Trafodion) Sent: Wednesday, February 11, 2015 11:54 AM To: Nick Dimiduk; [email protected]<mailto:[email protected]> Cc: Tapper, Gunnar Subject: RE: Phoenix monitoring Thanks Nick! Yes Data Services Manager (DSM) does provide the kind of perspective Pari is looking for but not for Phoenix. Actually, if Phoenix is providing metrics and information in a repository or via a REST interface, on queries, elapsed times, etc. then certainly it could be incorporated into DSM. Trafodion is doing that now in R1.0 and we are working on having DSM provide full support for Trafodion along with complete support for HBase (some of it being already there), in order to provide a comprehensive view of all workloads from HDFS, HBase, to SQL. We did have plans to provide a SDK so that data services like Phoenix could be easily plugged into DSM. Not a hard task – just time and effort ☺. If Pari is interested, we can certainly demo the tool and go from there based on interest. Rohit From: Nick Dimiduk [mailto:[email protected]<mailto:[email protected]>] Sent: Wednesday, February 11, 2015 9:01 AM To: [email protected]<mailto:[email protected]><mailto:[email protected]<mailto:[email protected]>> Cc: Jain, Rohit (Trafodion) Subject: Re: Phoenix monitoring Hi Pari, I'm not aware of a Phoenix-aware, end-to-end solution here. You can probably write a custom collector for a Phenox application that can report into OpenTSDB; I think JDBC metrics are available via JMX. That would be a great addition for that project! I've also seen a product demo from HP that does a lot of this for other tools on HBase, though I don't think it supports Phoenix yet (cc Rohit). Thanks, Nick On Wednesday, February 11, 2015, Pariksheet Barapatre <[email protected]<mailto:[email protected]><mailto:[email protected]<mailto:[email protected]>>> wrote: Hello All, Can you suggest good monitoring tool for phoenix. Our QA hbase cluster is crashing randomly. We wanted to know stats like which query causing the issue and how many queries running on that particular time window, resource consumption of query etc. Cheers Pari
