Before we jump into the queries, I would recommend that you have a look at this How-To Guide and listen to this Know How Network Call to understand more about these queries. As mentioned before, these have been posted by different clients and consultants so have a look and see if you had similar queries in your mind.
Queries posted by participants
AccAD 2.1 was scheduled for Ramp up in July ’08. Has the ramp up begun ?
Absolutely. The Ramp-up has begun. You can apply for a Ramp Up Program through your Account Executive.
Does the Client Front End (CFE) and Server Front End (SFE) need additional Hardware resources ?
You need Standard i386 hardware where a small Linux runs. So you would indeed need additional hardware, but is minimal.
What would be the value proposition to the clients by using AccAD ?
Please refer to this presentation and I believe that you would have a better understand to clearly state the value proposition of AccAD. I believe this presentation would also answer all your questions above: http://www.sdn.sap.com/irj/scn/go/portal/prtroot/docs/library/uuid/c06cefd0-5c71-2b10-b492-d4981b1b0f77
How are the TREX services (Index, Queue, and Preprocessor) organized in the Global Portal ? Are the indexes shared between all servers or are replicated ?
We have a TREX landscape for each portal landscape. We have the following configuration for Production :
There is one Queue server.
There are 2 index servers per physical server.
There is one preprocessor and nameserver per physical server.
There is an httpserver on all systems except a couple of them.
The index names are the same on each environment but the content is different as are the servers that make up the landscapes TREX environment.
Note : It would be better to include the SID in the index name. If the SID is included on TREX Server can be used for multiple Portals.
Are various services (Queue service, Index Service, crawler services etc) load balanced or is a particular server targeted for such services ?
There is a TREX installation. The installation is made up of the following processes:
1) Queue server
6) Alert server (optional).
You can configure the different components that make up TREX to run on different physical hosts but you cannot control how they are accessed that is the responsibility of the TREX System. Since we use a master slave configuration there is some failover available to us.
There is a KM crawler service. We configure the service to run a portal server not in the load balancer (we do this for all scheduled tasks). KM is configured to start the crawler after 10000 events or 30 minutes which ever comes first.
How are the KM repositories organized? Are they DB, FS? What were the considerations to support a specific decision ?
The primary repositories in the Corporate Portal Knowledge Management are DB repositories. There are pros and cons to each.
Some considerations are in the area of backup/restore. DB backup/restore strategy is much simpler however to restore single file in DB you must restore entire DB and connect to KM. FS restore of single file is possible. FS repositories rely on sync reports running in KM to ensure data does not get corrupt.
It is important not to modify files in FS repository from file system layer as this can cause huge inconsistencies. Performance also is a consideration. All options have different performance benefits/costs depending on usage (reads vs. writes).