4/26/20111Software as ServiceSoftware as ServiceBojan CukicWVUFor non-profit educational use only1WHY Software as Service?Product sales and license fee revenuesProduct sales and license fee revenues declining Users unhappy with high maintenance support fees Exceptions: product revenue for games continues to riseBli i2But on-line game revenues are growing too The ecosystem of Microsoft product maintains high revenue But service revenues now form ~7% of revenue.4/26/20112New pricing models Software as a service Access fee for any activity within the service suite Software ecosystems Free but not free Fee free services (Google, Yahoo!, Office Live) But advertisers pay for it3Trends44/26/20113Trends5Service Oriented Architectures64/26/20114Workflow7Web services stack84/26/20115Challenges for creating highly dependable service based systemsdependable service based systemsSean Banerjee1Hema Srikanth2Bojan Cukic1Lane Department of Computer Science and Electrical EngineeringWest Virginia UniversityMorgantown, WV, USAIBM, Lotus Division550 King StreetLittleton, MA, USAIntroduction Software as a Service (SaaS) is an emerging IT business model. Business case: Companies spend 70-80% of their IT budget in maintenance of infrastructure. SaaS improves operational efficiency by eliminating upfront investment and ongoingeliminating upfront investment and ongoing infrastructure maintenance. Services / Clouds trending up:• From $16B (‘08) to $42B (‘12) Growth comes with dependability expectations.4/26/20116OutlineExperiences with dependability analysisExperiences with dependability analysis of SaaS. How to get to “true dependability” of SaaS? Challenges in evaluating dependability of SaaS.Fundamental challenges in buildingFundamental challenges in building highly dependable SaaS applications. Middleman – The Network Service Provider Multi-tenancy – Security, Scalability and Availability ChallengesLog Files in Service Based Systems Access Logs Represents the outcome of all user transactions. p Significant “noise” for dependability analysis. Server Logs Monitors activities on the server. Begins reporting on start up, stops on server shut down. Business Process Logs Logs application critical requests and administrative tasks – user addition, deletion, etc. Customer Problem Reports Contains customer reported issues. Messages written in natural language. Merging of log files can enhance dependability analysis Merging logs is not trivial• Clock times, user information and failures need to be synchronized. Correlation with natural language reports a challenge.4/26/20117Challenges in Deploying Dependable SaaS Applications Middleman - Network Service Provider SaaS requires an active Internet connection On-premise application suites:• Client’s machine, middleware, application, local network• Impairments affect at most one organization. In SaaS, who is to blame for a network connection issue? • SLAs should include the user and the service provider.• How many network providers? Offline mode generally not be feasible in SaaS. • Storing data locally and synchronizing on availability may reduce data loss, but increases complexity. Multi-tenancy SaaS is multi-tenant Serving multiple customers through the same infrastructure. Not all the customers are created equal! First Degree Multi-tenancy: SalesForce.com Same software instance, components, database tables, serve all customers. Heavy reliance on threading. Second Degree Multi-tenancy: Intaact Replication with multiple instantiations of DB schemas offer a greater level of scalability. Low Level Multi-tenancy: Oracle services Unique cluster configuration for each client. Adds complexity for service provider!4/26/20118Multi-tenancy and Confidentiality, Integrity Data sensitivity: Collocation brings risks. Reputation fate sharing Clients who partake in illicit activities may tamper the reputation of others due to the use of shared resources. Replication increases reliability/availability, but iiikmay increase security risk too. A catch-22 situation. Risk analysis and mitigation. Many providers off-shore data stores. Legal assurances and ramifications?Multi-tenancy and Scalability Performance should not degrade with addition of new clients. For SaaS, scalability on demand is a challenge. Complexity of multi tenancy impacts solution approaches. The illusion of infinite resources. Scaling up should not require downtime•Scaling up should not require downtime. Distributing the application across multiple servers, virtual machines. Highly dynamic environments. • Imply issues with data integrity and security.4/26/20119Multi-tenancy and AvailabilityRequests per day Requests per hour q p yof weekq pof day• SaaS customers access the services from all across the world. • Workload pattern appears remarkably different the literature for traditional web servers wherein peaks are noticed during week days and work hours. •Goseva‐Popstojanova [2004, 2006], Tian [2004], Faber [2006]• 24/7 usage means a low tolerance of downtime. Summary 24/7 nature of SaaS Customer and provider teams are global. Highly dependable service systems: What is high enough? Aspects of dependability compete with each other. Need to balance dependability goals. Selection of adequate metrics for SaaS SLA remains an open problem. Current ones are simple and may not be impartial. The role of networks in
View Full Document
Unlocking...