Adobe publishes Capacity Planning Guides for select groups of LiveCycle components and Solution Accelerators. These currently are:
– Interactive Statements (Solution Accelerator)
– Forms ES2 and Reader Extensions ES2 (combined)
– Output ES2
– PDF Generator and Assembler (combined)
– Rights Management ES2
– Digital Signatures ES2
– Process Management ES2
Because of the sensitive nature of the information in these, they are only available to partners and those customers who have signed a Non-Disclosure Agreement (NDA) with Adobe. If you belong to this group but has not received these Guides, please contact your Adobe Partner or Adobe Account Manager.
These guides are updated for major releases of LiveCycle. They are prepared based on tests conducted on servers in Adobe’s Server Performance Lab. Since it is practically impossible for Adobe to conduct tests for LiveCycle on all variations of all types of servers offered by major manufacturers such as HP, Dell, IBM, Sun/Oracle and Fujitsu, a server performance index is required to compare the relative performance of one server against that of another.
Such an index should be as sensitive as possible to the transaction throughput of LiveCycle orchestrations, short-lived as well as long-lived. This “sensitivity” can be quantified using the Correlation Coefficient of LiveCycle transaction throughput to the index. Based on various tests, Adobe has chosen the Standard Performance Evaluation Corporation’s (SPEC) CINT2006 as the index with the best correlation coefficient to LiveCycle transaction throughput performance.
CINT2006 is the integer (as opposed to the floating point CFP2006) component of their CPU2006 benchmark. Specifically, it is the SPEC CINT2006 Base Rate (“Throughput: SPECint_rate2006″) that should be used.
To find your server’s SPEC INT2006 Throughput Base Rate, go to the CPU2006 Search Form and search by various criteria such as ‘System’, ‘Hardware Vendor’ etc. For example, choose ‘System’, type in the search term “Proliant DL365″ and click the button ‘Execute Simple Fetch’. In the results page (shown below), go to the ‘CINT2006 Rates‘ section (not the ‘CINT2006′ section) and identify the system that is closest to yours and look at its ‘Baseline‘ score.
In the above example, the CINT2006 Baseline Rate for an HP Proliant DL365 G5 with a single 2.3 GHz AMD Opteron 2356 quad-core CPU is 45.7
Gardner Buchanan is the primary author of most of the LiveCycle Capacity Planning Guides. These Guides have an appendix where he describes how to use the SPEC CINT2006 benchmark of your hardware to predict how LiveCycle would perform on it. The following is excerpted:
Occasionally an exact match for a system cannot be found within the published SPEC results. There are some creative ways to guess an approximate value based on information that is available, for example:
• If the available results are for a different number of CPU cores that you have, you might linearly scale the reported value to the actual number of CPU cores you have.
• If the available results are for a different CPU speed, you might linearly interpolate the required CPU speed between speeds reported in the benchmarks.
• If there are no results for a particular manufacturer or model, you might find a similar or identical model from a different manufacturer, for example a certain UltraSPARC CPU type might be reported by Fujitsu Computer Systems, but not by Sun Microsystems.
Please note that earlier versions of LiveCycle Capacity Planning Guides used the older SPEC CINT2000 benchmark. SPEC has since retired it.