Scaling of Apigee components

[#1]

Hello All,

.

We would like to scale zookeeper, cassandra and postgress.

We have a traffic of 30 TPS, Also multiple child tables, which is consuming the postgres space.

Today, we have postgress in master (active)/Slave (inactive) configuration. Both master and slave as 1TB memory each.



Do we need to scale both master and slave ?

Or can we take analytics data from postgress server to some other back up location. If yes, how to show that analytics data in UI by restoring the specific period data?

Regards,

Ch.Venkat.

[paulmibus]

You will need to scale both the primary and the standby database disk space, since both will hold identical data due to replication.

Backing and up and restoring specific data would require exporting and removing child fact tables in the analytics database. The purge process is detailed at http://docs.apigee.com/api-services/latest/recurring-analytics-services-maintenance-tasks. You could take the built in script and edit it to dump the data to a file prior to purging, then reload that file at a later time or into a different database for later analysis.

It is worth noting that when you prune old detailed analytics data, your summary report data is kept in the database and will continue to be accessible. You only lose the ability to look at detailed, per-transaction data.

[#3]

Hello Paul,

I'm more interested in the custom reports created by us. we would like to take backup of custom reports data and store in some other disk before purging.

Later, need to get those files to visible in my UI in the present reports (however should show the specific period which i access and that period exists in my restored data)

Hope i'm clear.

Assume i have taken backup of Jan-2016 custom reports data and stored in someother disk. In february my report shows only feb data.

if i restore the Jan data to present disk then it should show even Jan data along with Feb in my present reports by selecting the jan in date tab.

Bottom Line: I would like to see the historical data whenever it's needed by restoring the specific period.

How to achieve it?

Regards,

Ch.Venkat.

[paulmibus]

Hi Venkat,

The process I described previously would work for your use case. Custom reports draw on the detailed, per-transaction data stored in the analytics database fact tables. By exporting that data for later use, you could preserve the data needed to generate those reports. You can see all available fact tables in the analytics database on an analytics database host:

psql -h /opt/apigee/var/run/apigee-postgresql -U apigee

\dt analytics.*

Child fact table partitions take the form analytics."<org>.<env>.fact_###".