Splitting APS:
Adaptive Processing Server has approx. 23 common services, running all services together keeps deployment more simple and it also keeps lower RAM requirement but it does not have better system scalability. To get better system scalability we split the APS into multiple instances it enables better resource uses and isolation specific tasks into java container and each java container would have its own java heap setting. It costs greater RAM uses when system is Idle as well as system is busy but in return it has better system scalability.
Adaptive Processing Server runs approx. 23 processes with 1 Giga byte of java heap size
This is the default that APS does.
RAM usage can be increased when splitting the APS but in return it gives better system scalability.
Please keep in mind when splitting APS. It requires at least 16G or more of RAM per BI cluster
It can be 32 or 64 GB of RAM that would be more beneficial.
Some benefits includes:
- BI Platform will have better scalability and reliability. APS Splitting is a good idea if you are reporting over BW.
- System Resources can be utilized more better particularly RAM for better system scalability.
- Server side process threads execution will be more efficient.
- System can support many concurrent users with less downtime.
Services in each splitted servers:
1. DSL Bridge Services: is responsible for querying and serving data to report
- DSL Bridge Service
- Web Intelligence (Interactive Analysis) Monitoring Service
- Security Token Service
- Tracelog Service(Included by default)
Recommended heap size: 4-8g
2. Search Service: User can search keyword into BI Launchpad.
- Platform Search Service
- Tracelog Service(Included by default)
Recommended heap size: 2-3g
3. OLAP Services: These enable Webi analysis reporting tool and BEx web application service
- Multi Dimensional Analysis Service
- BEx Web Applications Service
- Tracelog (Included by default)
Recommend heap size: 4-8
4. LCM Services : These handle promotion management and version control
- Promotion(Life Cycle) Management Clear Cache Service
- Promotion(Life Cycle) Management Service
- Visual Difference Service
- Tracelog (Included by default)
Recommended heap size: 1-2g
Note: Use one Promotion Management Service per BI clusters because overrides database does not support multiple Promotion instances.
5. Publishing Services: These handles personalizing reporting publication
- Publishing Post Processing Service
- Publishing Service
- Tracelog (Included by default)
Recommended heap size: 2-4g
6. Web Intelligence Services
- Visualization Service
- Rebean service
- Web Intelligence (Interactive Analysis) Monitoring Service
- Document Recovery Service
- Tracelog (Included by default)
Recommended heap Size: 2-4g
7.Others services
- Adaptive Concavity Service
- Custom Data Access Service
- Excel Data Access Service
- Client Auditing Proxy Service
- Monitoring Service
- Security Token Service
- Translation Service
- Insight to Action Service
- Data Federation Service
- Tracelog (Included by default)
Recommended heap Size: 2-4g
Steps:
1. Stop and disable APS
2. Right click and uncheck automatically start server.
3. Right click over APS select “Clone Server”, rename it (Adaptive Processing services1 (DSL Bridge)) and then “Save and Close”.
4 .Right click over newly cloned server and select “Select Services”.
5. Remove all services except DSL Bridge Service and Security Token Service.
6. Right click select “Properties” go to command line change –xmx1g to –xmx4g and then mark check on automatically start servers option.
Best way to do that select all and then copy and paste to notepad edit and then copy and paste to command line.
7. Start and then Enable the newly splitted server.
8. Please follow same steps to do rest.
Please note that you do not start and enable Adaptive Processing server and should not start automatically when SIA restarts..
Thanks,
Imran