SAL A Service Level Agreement Language Donna Eng Dillenberger engd us ibm com October 22 2003 Document Changes Date 9 22 03 10 08 03 10 22 03 Changes Document Created Included variables control flow and functions Included Language Reference Manual Contents An Introduction to SAL 4 Tutorial 10 A First Example 10 More Examples 13 Reference Manual 16 Lexical Conventions 16 Appendix A 22 Software Development Environment 23 Bibliography 24 An Introduction to SAL The SAL language is designed to help Information Technology IT professionals and Business Process Modelers quickly build metrics prototype evaluate and optimize the configuration of their systems using Service Level Agreements SLAs Service Level Agreements are legal contracts used by business organizations to specify levels of performance availability storage server and network capacity that an IT organization will provide to support the business processes When these levels are not met IT organizations are penalized by compensating the business in dollar amounts for the time that lack of access to the computer infrastructure causes business transactions to be missed Examples of Service Level Agreements rules are This IT organization will support the following 4 Business Processes 1 Trading Application 2 Prospectus Browsing Application 3 Risk Portfolio Analysis 4 Maintenance Background jobs The Trading Application 1 Between 9 am and 5pm will have a Mean Time to Recovery 1 of 2 seconds 2 All requests to the Trading Application will receive a response within 3 seconds 3 The profit from each Trading Application request is ten dollars for trades 4 The IT cost of each Trading Application request should not exceed one dollar 5 If the response time is greater than 3 seconds and it s after 9 am then a penalty of 100 dollars shall be incurred for each transaction The Prospectus Browsing Application 1 Between 9am and 10 pm will have a Mean Time to Recovery of 5 minutes 2 All requests to this Application will receive a response within 1 minute 3 The profit from each request for this application is two dollars 4 The IT cost of each request to this application should not one dollar 5 No penalty to IT if the above rules are not met for Prospectus Browsers The above rules are best effort 1 Mean time to recovery entails that this application must not be inaccessible for longer than 2 seconds The Risk Portfolio Analysis Application 1 Between 6pm and 12 pm will have a Mean Time to Recovery of 2 hours 2 All requests to this Application will receive a response within 8 hours 3 The profit from each request for this application is two hundred dollars 4 The cost of each request to this application should not exceed 10 dollars 5 If any of the above rules are missed IT s payment to the business will be 100 dollars for Each transaction missed because Mean Time to Recovery was not met between 6 pm and 12 pm Each transaction that failed to get a response within 8 hours Maintenance and Background work 1 Will have a Mean Time to Recovery of 4 hours between 1 am and 7 am 2 All requests to this Application will receive a response within 24 hours 3 The profit from each request for this application is 3 dollars 4 The cost of each request to this application should not exceed 2 dollars 5 No penalty to IT if the above rules are not met for Prospectus Browsers The above rules are best effort To specify the above using SAL one would write Application Trading For Scope 900 to 1700 MeanTimeToRecovery 2 seconds ResponseTime 3 seconds Profit 10 dollars TxCost 1 dollar For Scope 900 ResponseTime 3 seconds Penalty 100 dollars Application ProspectusBrowsing For Scope 900 to 2200 MeanTimeToRecovery 5 minutes ResponseTime 1 minute Profit 2 dollars TxCost 1 dollar Penalty 0 dollars Application RiskPortfolioAnalysis For Scope 600 to 1200 MeanTimeToRecovery 2 hours ResponseTime 8 hours Profit 200 dollars TxCost 10 dollars For Scope 600 1200 ResponseTime 8 hours MeanTimeToRecovery 2 hours Penalty 100 dollars Application MaintenanceBackground For Scope 100 to 700 MeanTimeToRecovery 4 hours ResponseTime 5 hours Profit 3 dollars TxCost 2 dollars Penalty 0 dollars The compiler for SAL will Check that for each defined Application there are MeanTimetoRecovery ResponseTime Profit TxCost and Penalty values defined and specified in the right units Transform the Application definitions above to java classes The java code could then be fed into SLA deployment machines these don t currently exist yet that take the above definitions and 1 Creates Monitoring Agents for each Application and measures whether the above thresholds are violated 2 In more sophisticated SLA deployment machines forecast when the defined thresholds will be violated Parameter passed into Forecast would request how far into the future a forecast is desired 3 Provisions deprovisions reconfigures IT routers servers and storage to prevent Threshold violation from occurring 4 In more sophisticated SLA deployment machines the IT adaptation would maximize profit minimize cost within the boundaries of not incurring IT penalties The programmer would invoke the above actions 2 4 by coding Actions Forecast Optimize Provsion For the above the compiler would derive calculate the right parameters to invoke the above functions Forecast Optimize Provision The implementations for 1 4 above are meant to be pluggable Different vendors can supply different implementations to differentiate themselves The interfaces will be generated by the compiler but the package implementations will not be provided these would be vendor specific This would print the results of Creating Agents Forecast warnings what provisioning configuration actions were taken and what optimization tradeoffs were invoked A key initial step in writing a SLA is to identify the Business Processes the IT infrastructure will support IT implementations should be driven by the business needs for the measurement of progress and fulfillment of business goals e g ROI EBIT At a fundamental level each organization has a general goal e g maximize Return On Investment ROI or Earnings Before Interest and Taxes EBIT An organization s next step is to translate those goals into Lines of Businesses that can support maximizing these business goals For example a Financial Institution may choose to offer three Lines of Businesses to maximize ROI A Trading Service A Risk Portfolio Analysis Service A Prospectus Downloading Service These three types of services all would have quarter profit
View Full Document
Unlocking...