-
What is the hardware requirement for ECC?
ECC is a data driven application, and requires large number of resources. Hardware Requirement for 15 datasets each with 2 million of records (total 30 millions of records), 50 named users):
Core: 2
Memory: 32 GB
Disk Space: 80 GBThe Note 2495053.1 provides the install information, sizing guidelines and also the details of the setup and configuration. This note is a source of reference.
-
What is the recommended number of Data Loads that can be run in parallel without impacting server performance.
ECC is tunable to handle different resources, data sizes and user-base. So for example if customer runs Data load at midnight where dashboard load is very low they can use all ECC machine threads for data load. As well if the ECC machine has 8 cores and every job is configured to use 4 workers (another configuration) then if the customer configures 2 parallel jobs that means 8 threads will be used in ETL and accessing dashboards by end users will be affected. Also it depends on how long each job takes. So in general everything is configurable. Try as much as it is feasible to avoid busy time for running ETL, and if they want to run more than one job at the same time need to make sure that have the right cores to handle all jobs and at the same time to handle end user dashboard requests. -
What is the maximum volume consideration for a single data set.
ECC supports replication but not partitioning. Datasets can be distributed but the single dataset needs to be completely hosted in one machine (it may be replicated completely to other machines). To get the right performance the customer needs to have good machine resources to handle big data. For example, the average document/record size is 1.5KB. So 10M records are of size around 15 GB in desk. To get good performance most of this needs to be in memory at the same time, so machine with 32 GB of RAM should be good to handle that. If there are more records, it will need to update RAM. Same for CPU, more CPU helps on more threads to work on the records. -
Are there any locks during Data loads on ECC or EBS side.
No locks. But new data will be visible to end user at the end of the load. If there are currently 1000 records, and starts incremental load for this dataset which adds more 500 records, the end users will keep working with the 1000 records until the job gets finished, when they can see the new 500 records.