1. Jožef Stefan Institute Oct. 8-9, 2005ACS 2005 Workshop Comparison of EPICS and ACS M. Plesko – JSI and Cosylab, in collaboration with Phil Duval, DESY

2. Jožef Stefan Institute www.cosylab.com M.Plesko, compare EPICS and ACS2 History of This Presentation Started at PCaPAC2002 in Rome –Made survey on free CS from the community Compare features Discuss Excellent feedback, but.... Difficult to compare: –Each author responded to a long poll Updated survey during ICALEPCS 2003

3. Jožef Stefan Institute www.cosylab.com M.Plesko, compare EPICS and ACS3 Evaluated Control Systems Complete systems –ACS –COACK –DOOCS –EPICS –TACO/TANGO –TINE APIs and GUIs –Abeans and CosyBeans –ACOP –CDEV Commercial –IFix –VisPro –Vsystem

4. Jožef Stefan Institute www.cosylab.com M.Plesko, compare EPICS and ACS4 The Layers of Control System Packages EPICS DOOCS COACK CDEV TINE chan. acc. ACS driver data SERVER commu- nication API visual support CLIENT ACOP AbeansCosyBeans TANGO MEDM JDM DDD Applications (alarm manager, GUI, logger, trending, scripting etc.): while all have them, their quality, flexibility, configurable etc. makes the largest difference

5. Jožef Stefan Institute www.cosylab.com M.Plesko, compare EPICS and ACS5 CS Comparison: Poster WP536

6. Jožef Stefan Institute www.cosylab.com M.Plesko, compare EPICS and ACS6 CS Comparison: I/O Integration

7. Jožef Stefan Institute www.cosylab.com M.Plesko, compare EPICS and ACS7 CS Comparison: Application Development

8. Jožef Stefan Institute www.cosylab.com M.Plesko, compare EPICS and ACS8 The Really Difficult Questions Applications developers: –How easy is it to learn the system? –How easy is it to create an application? –What features are missing? –What features are good? –What are the limitations? –Are the applications primitive (connecting data to display tools only)? –Can the applications be programmed? Operators: –How good are the applications in general? –What's missing? –What could be better? –What is the information density on a typical application? –How quickly can I find and isolate problems? –How often do failures in the control system hinder operations?

9. Jožef Stefan Institute www.cosylab.com M.Plesko, compare EPICS and ACS9 Other Difficult Questions Engineers –How easy is it to remotely test my equipment? –How easy is it to find problems? –How good is the alarm system notification of pending hardware failures? Inventors of the CS –what’s your problem? Everything is perfect and there is nothing better! Physicists –How good are the on-line analysis tools? –How easy is it to get the live data (archived data) into my favorite analysis tool? Directors of institutes –Could machine operations be better/smoother/less expensive with another system? –mostly care about PR

10. Jožef Stefan Institute www.cosylab.com M.Plesko, compare EPICS and ACS10 Difference in Device Interfaces client server 1 master interface EPICS is “Narrow” CA_put(“device”,data) ACS is “Wide” client server 1 interface/device dataOut=device.put(dataIn)

11. Jožef Stefan Institute www.cosylab.com M.Plesko, compare EPICS and ACS11 Problems Mapping EPICS and ACS Runtime database types –EPICS: predefined –ACS: no runtime database Process Variables –EPICS PV are primitive types –ACS RW/RO types are objects –but ACS CDB data are primitive types Devices –EPICS uses naming convention –ACS uses IDL Commands –EPICS uses variables to trigger commands –ACS uses methods

12. Jožef Stefan Institute www.cosylab.com M.Plesko, compare EPICS and ACS12 Mixing Orange Clients with Apple Servers 1.‘Apple-to-orange’ gateway a separate process utilizing the client/server APIs of both systems 2.Client-side ‘apple-plugs’ client program developers think they are talking to oranges, but they are really speaking native ‘apple’ 3.Server-side ‘orange-translators server IOCs think they are being addressed by apples but are really speaking native orange.

13. Jožef Stefan Institute www.cosylab.com M.Plesko, compare EPICS and ACS13 Why Translators are Better Server-side systematics are guaranteed to be there. –local alarm server, local history server, queries, etc. The data are as close to the source as possible can use the best applications for each problem Gateways bring a set of intermediate problems –e.g. connectivity problems are difficult to locate –another link in the chain Plugs could miss functionality

14. Jožef Stefan Institute www.cosylab.com M.Plesko, compare EPICS and ACS14 EPICS, TINE and DOOCS Translator: a Model for ACS-EPICS Communication run EPICS2TINE directly on the IOC –don't speak channel access at all and access EPICS database directly DOOCS servers and clients offer the traditional SUN RPC interface as well as a TINE interface –EPICS IOCs are immediately available to DOOCS DDD clients. Using TINE2EPICS, the DOOCS IOCs are likewise available to EPICS MEDM clients. Estimate from Vista: it takes them 2-4 weeks/tool to integrate their tools into any system

15. Jožef Stefan Institute www.cosylab.com M.Plesko, compare EPICS and ACS15 EPICS PV as ACS Property enum ChannelType { STRING, INT, SHORT, FLOAT, ENUM, CHAR, LONG, DOUBLE }; interface EPICSAccess : ACS::CharacteristicComponent { ACS::Property createPropertyForChannel(in string name, in ChannelType type); };

16. Jožef Stefan Institute www.cosylab.com M.Plesko, compare EPICS and ACS16 Conclusions With translators and plugs we can mix any set of apples and oranges Appeal to CS package developers –think more about services that could be used in a generic way by other control systems –provide good general purpose applications and tools –good documentation and tutorials won’t hurt either