Hi all, Had an offline discussion with Nuwandi and following are some points.
- We can filter region from IP and provide ability to drill down, but if we follow this approach, region is not shown in the tables. This leads to a different user experience. Since we don't have any other possible solutions as suggested by Mohan, let's go with the first approach. Please create a jira for this and state required changes to the stream definitions. On Wed, Jun 29, 2016 at 12:38 PM, Mohanadarshan Vivekanandalingam < [email protected]> wrote: > HI Johann & Hasintha, > > Appreciate any feedback on above, then we can implement accordingly and do > some perf tests.. If drill capability is required for Map then we need to > follow above option.. Here, we can > > 1) Change the stream in Identity level and publish > 2) Change at DAS level by adding another attributes in Siddhi scope.. > > But, I like to avoid option [2] to avoid unnecessary event conversions > (which impact the TPS).. Then, option [1] is the only solution in this > case.. > > Thanks, > Mohan > > > On Tue, Jun 28, 2016 at 4:16 PM, Nuwandi Wickramasinghe <[email protected] > > wrote: > >> Hi all, >> >> IS Analytics dashboard contains a map in which we have marked the regions >> of users who have authenticated through Identity Server. Our concern is >> whether we should allow to drill down from this map. >> >> Normally when drilling down from a gadget, in the dashboard we only show >> the data related to that selected field. And there's a data table below >> which shows the raw data corresponding to the selected field. This data >> table is populated using the raw stream which is published from Identity >> Server side. >> >> When implementing drill down capabilities from the area chart (map), we >> are facing an issue in populating the data table since the raw data stream >> published from IS contains only the remoteIp (which is used to determine >> the region) but not the region field itself. >> >> Regions and IP addresses have one to many relationship. Which means a >> large range of IPs corresponds to one region. Therefore we can get the >> region from IP but not vise versa. This leaves the issue that we cannot >> search in the raw stream using the region selected from the map. >> >> Suggested solution is to change the raw data stream to contain an >> additional 'region' attribute which could be used to search for the data >> related to a particular region. The value of region for the raw data will >> be sent as 'NOT_AVAILABLE'. We are planning to update this value to the >> correct region according to the remoteIp value using a spark query. We >> considered updating the records from the execution plan but that could >> delay the real time process. >> >> If we are going to implement this, Hasintha will have to introduce the >> new region attribute to the stream with a default value. Also this might >> introduce some performance degrade in the db level. >> >> We are still considering the need of drilling down from the region map. >> Appreciate other opinions regarding the subject. >> >> Thanks >> Nuwandi >> >> -- >> >> Best Regards, >> >> Nuwandi Wickramasinghe >> >> Software Engineer >> >> WSO2 Inc. >> >> Web : http://wso2.com >> >> Mobile : 0719214873 >> > > > > -- > *V. Mohanadarshan* > *Associate Tech Lead,* > *Data Technologies Team,* > *WSO2, Inc. http://wso2.com <http://wso2.com> * > *lean.enterprise.middleware.* > > email: [email protected] > phone:(+94) 771117673 > -- Hasintha Indrajee WSO2, Inc. Mobile:+94 771892453
_______________________________________________ Dev mailing list [email protected] http://wso2.org/cgi-bin/mailman/listinfo/dev
