RailCloud: A Reliable PaaS Cloud for Railway Applications Bijun Li , - PowerPoint PPT Presentation

Platzhalter für Bild, Bild auf Titelfolie hinter das Logo einsetzen RailCloud: A Reliable PaaS Cloud for Railway Applications Bijun Li , Rüdiger Kapitza TU Braunschweig 06.10.2016 This work is supported by Siemens international Rail Automation Graduate School (iRAGS)

RailCloud PaaS Cloud for Railway Applications  A PaaS cloud for railway applications  Shared by small and medium-sized transportation companies  Reliability and safety guarantee 06.10.2016 | Bijun Li | RailCloud: A Reliable PaaS Cloud for Railway Applications | Page 2

PaaS Cloud in a Nutshell What is a PaaS Cloud?  A computing platform for software development delivered over the Internet How to use it?  Software developers can quickly deploy Applications applications, without infrastructure management tasks Platform (OS, Middleware, Runtime) Existing PaaS Clouds? Storage Networking Servers Virtualization  Google App Engine (GAE), Microsoft Azure, OpenShift, Cloud Foundry etc.  Recent evolvement with container technology 06.10.2016 | Bijun Li | RailCloud: A Reliable PaaS Cloud for Railway Applications | Page 3

Existing PaaS Clouds? Reliability? Horizontal Scalability  Load Usually for load balancing Balancer  Tolerate crash-stop failures Goal of RailCloud  Mostly for stateless applications  App App App App Easy deployment of replicated stateful applications with automatic coordination to guarantee reliability  Easy deployment of legacy railway applications Issues  Lack of support for replicated stateful applications  Cannot tolerate malicious attackers  Complex deployment and coordination for cloud customers 06.10.2016 | Bijun Li | RailCloud: A Reliable PaaS Cloud for Railway Applications | Page 4

Outline  Reliability in PaaS Clouds  RailCloud Design  Byzantine Fault-Tolerant Applications in the Cloud  Trusted Proxy: Making Replicated Systems Transparent  Conclusion and Future Work 06.10.2016 | Bijun Li | RailCloud: A Reliable PaaS Cloud for Railway Applications | Page 5

RailCloud Architecture Client 1 Client 2 RailCloud Trusted Trusted Trusted Trusted Proxy Proxy Proxy Proxy BFT BFT BFT BFT Replica Replica Replica Replica APP APP APP APP Replica Replica Replica Replica Developer Host 4 Host 2 Host 3 Host 1 Request/response Flow Built-in Service of RailCloud BFT Message Exchanges Application Deployment 06.10.2016 | Bijun Li | RailCloud: A Reliable PaaS Cloud for Railway Applications | Page 6

Byzantine Fault Tolerance in the Cloud Byzantine Fault-Tolerance (BFT) Protocols  Tolerate crash-stop failures and arbitrary and malicious behaviors Req Req Req Rep Req  3 f +1 replicas to tolerate f faults  Rep Message exchanges for agreement Rep Integration of BFT into Cloud Infrastructures  Infrastructure level: Depsky, Fitch, TClouds etc.  Middleware level: Thema etc. RailCloud: PaaS Level + Automatic Deployment Extension 06.10.2016 | Bijun Li | RailCloud: A Reliable PaaS Cloud for Railway Applications | Page 7

Base of RailCloud OpenShift Origin v3 OpenShift API Server  Kube Proxy Docker container Kube Proxy Build Deployment packaging POD POD Controller Controller Kube Proxy POD POD OpenShift Master Container Container POD POD  Kubernetes container Container Container Developers Docker Kubernetes Container Container cluster management Scheduler API Server Docker Kubelet Docker Kube Replication Kubelet Controller Controller  Application lifecycle Kubelet Kubernetes Master management MASTER(S) NODE(S) 06.10.2016 | Bijun Li | RailCloud: A Reliable PaaS Cloud for Railway Applications | Page 8

Implementation and Deployment BFT Service Layer  BFT image (BFT-SMaRt) BFT Service  BFT pods BFT Service BFT Service BFT Service POD  BFT services POD POD POD BFT Container BFT Container BFT Container BFT Container Application Deployment App Service App Service Networking App Service App Service POD POD  POD Connect each BFT service POD App Container App Container to application service App Container App Container  Expose BFT services RailCloud 06.10.2016 | Bijun Li | RailCloud: A Reliable PaaS Cloud for Railway Applications | Page 9

Trusted Proxy: Making Replicated Systems Transparent Why transparent? Minimum modifications to clients  HTTPS connections  Web-based railway applications  Implements client-side BFT library Friendly to low-bandwidth clients  No redundant requests/replies Hide details of replicated system  Simple and secure interface to clients Throughput improvement 06.10.2016 | Bijun Li | RailCloud: A Reliable PaaS Cloud for Railway Applications | Page 10

Trusted Proxy in RailCloud Message Flow Client Client Read Optimization Secure connection Secure connection Connection Handler Connection Handler Connection Handler Connection Handler En/decrypt En/decrypt En/decrypt En/decrypt Reply vote Reply vote Reply vote Reply vote (Fast read cache) (Fast read cache) (Fast read cache) (Fast read cache) Trusted Proxy Trusted Proxy Trusted Proxy Trusted Proxy Connection Handler Connection Handler Connection Handler Connection Handler Ordering + Ordering + Ordering + Ordering + Execution Execution Execution Execution Host 3 Host 4 Host 1 Host 2 06.10.2016 | Bijun Li | RailCloud: A Reliable PaaS Cloud for Railway Applications | Page 11

Outline  Reliability in PaaS Clouds  RailCloud Design  Byzantine Fault-Tolerant Applications in the Cloud  Trusted Proxy: Making Replicated Systems Transparent  Conclusion and Future Work 06.10.2016 | Bijun Li | RailCloud: A Reliable PaaS Cloud for Railway Applications | Page 12

Conclusion and Future Work RailCloud  Integrate BFT for reliability demands into PaaS Cloud  Easy deployment of BFT applications  Automatic coordination among replicated services  Using trusted proxy to make replicated system transparent Future Work  Explore more functions of trusted proxy 06.10.2016 | Bijun Li | RailCloud: A Reliable PaaS Cloud for Railway Applications | Page 13

Appendix Related Works  Bessani, A., Correia, M., Quaresma, B., Andre, F., Sousa, P.: Depsky: dependable and secure storage in a cloud-of-clouds . ACM Transactions on Storage (TOS) 9(4), 12 (2013)  Cogo, V.V., Nogueira, A., Sousa, J., Pasin, M., Reiser, H.P., Bessani, A.: Fitch: Supporting adaptive replicated services in the cloud . In: DAIS’13  Garraghan, P., Townend, P., Xu, J.: Using byzantine fault-tolerance to improve dependability in federated cloud computing . International Journal of Software and Informatics 7(2), 221 – 237 (2013)  Verissimo, P., Bessani, A., Pasin, M.: The tclouds architecture: Open and resilient cloud-of-clouds computing . In: Dependable Systems and Networks Workshops (DSN-W), 2012 IEEE/IFIP 42nd International Conference on. pp. 1 – 6. IEEE(2012)  Merideth, M.G., Iyengar, A., Mikalsen, T., Tai, S., Rouvellou, I., Narasimhan, P.: Thema: Byzantine-fault-tolerant middleware for web-service applications . In: Reliable Distributed Systems (SRDS), 2005 24th IEEE Symposium on. pp. 131 – 140. IEEE (2005) 06.10.2016 | Bijun Li | RailCloud: A Reliable PaaS Cloud for Railway Applications | Page 14