Table of Architecture and Sizing Scenarios
Use the following table to identify an applicable architecture. Use the preceding sections to compute your implementation’s Data Volume Class (A-E+, from section above) and Concurrent User Load Impact (1-500+, from section above).
Server type keys
* 1 vCPU = 1 virtual processor identifiable by the operating system.
** The heap sizes are total heap on a single server. We recommend that you have multiple JVM instances, each with 8 GB heap. However, this approach requires a load balancer. For more information, see Horizontal Scalability (Scaling Out) and Java Instance Efficiency and Heap Size in Clustering and Horizontal and Vertical Scalability.
*** The heap sizes are total heap on a single server. We recommend that you have multiple JVM instances, each with 16 GB heap. However, this approach requires a load balancer. For more information, see Horizontal Scalability (Scaling Out) and Java Instance Efficiency and Heap Size in Clustering and Horizontal and Vertical Scalability.
**** You can use a Combined role server to deploy both the application and the database server provided performance is not a concern. EcoSys does not recommend a combined role server for Production deployment.
Further recommendations and assumptions
-
Memory sizes for Java heap do not include RAM needed for underlying operating system, which is typically ~4 GB additional physical memory.
-
Application servers require at least 50 GB disk free for log files and caches.
-
All servers should be 64-bit architectures and 64-bit operating systems and applications.
-
CPUs are sized for modern (last 3 years) clock speed and local cache sizes.
-
Network hardware requires to support gigabit speeds.
-
For database-intensive deployments, dedicated storage for the database is recommended. RAID-10 is preferred for performance with latency < 10ms.
-
This table does not include the integration server sizing. Contact EcoSys support for sizing recommendations of Integration Server.
-
This table does not include server sizing for any Disaster Recovery environment.