Caveats on Accuracy in Sizing
-
EcoSys has a comprehensive sizing and performance tuning methodology that is followed after the requirements phase of the implementation. We recommend that the initial sizing estimates are reviewed and validated during the implementation after key inputs on data volume and usage profiles by role are designed.
-
We recommend that the customer undertake performance testing on the target environment using typical usage scenarios by role, prior to rolling out the software to a large group of users. EcoSys support can assist with this exercise.
-
Shared resources introduce external variables that affect performance. These include: shared storage, database clusters, networks, and especially virtualized environments.
Concurrent Users, Data Volume, and Assumptions
Guidelines for estimating concurrent user load:
-
If no additional information is available, a general guideline is 1 concurrent user for every 3-5 named users.
-
Lightweight or read-only users may be counted as ½ of a concurrent user each.
-
Automated integration interfaces may be counted as 2-3 concurrent users each.
Guidelines for estimating data volume:
-
Most implementations can use the "Typical" data volume category below.
-
Implementations with very deep project structures (more than 20,000 WBS elements per project), a large number of projects (more than 100,000 projects), or a large number of cost line items (more than 10MM transactions per month) may use the "Large" data volume category below.
Assumptions:
-
All platforms are assumed to be 64-bit and have 1-Gbit networks connecting the application servers and database servers.
-
A virtual processor is defined as a single CPU core or virtual processor, as identified as an available processor to the application layer.
-
Memory sizes for App Server do not include RAM needed for underlying operating system, which is typically ~4 GB additional physical memory.
Infrastructure Scenario Recommendations
|
Named Users |
Small Data Volumes |
Large Data Volumes |
|||
|---|---|---|---|---|---|
|
App |
DB |
App |
DB |
||
|
|
20 GB mem 4 vCPUs 50 GB disk |
12 GB mem 4 vCPUs 50 GB disk |
24 GB mem 4 vCPUs 50 GB disk |
16 GB mem 4 vCPUs 100 GB disk |
|
|
|
32 GB mem 8 vCPUs 50 GB disk |
16 GB mem 8 vCPUs 100 GB disk |
36 GB mem 8 vCPUs 50 GB disk |
24 GB mem 8 vCPUs 150 GB disk |
|
|
|
64 GB mem 8 vCPUs 50 GB disk |
24 GB mem 8 vCPUs 100 GB disk |
2 @ 48 GB mem 2 @ 8 vCPUs 2 @ 50 GB disk |
36 GB mem 8 vCPUs 150 GB disk |
|
|
|
2 @ 48 GB mem 2 @ 8 vCPUs 2 @ 50 GB disk |
36 GB mem 8 vCPUs 150 GB disk |
2 @ 64 GB mem 2 @ 8 vCPUs 2 @ 50 GB disk |
64 GB mem 12 vCPUs 250 GB disk |
|
|
|
2 @ 64 GB mem 2 @ 8 vCPUs 2 @ 50 GB disk |
48 GB mem 12 vCPUs 250 GB disk |
3 @ 64 GB mem 3 @ 8 vCPUs 3 @ 50 GB disk |
96 GB mem 16 vCPUs 250 GB disk |
|
|
> 500 |
A customized sizing exercise is recommended for configurations of this size. |
||||