1. Distributed Systems Lecture 1 Introduction to distributed systems 1

2. Distributed systems “A collection of (probably heterogeneous) automata whose distribution is transparent to the user so that the system appears as one local machine. This is in contrast to a network, where the user is aware that there are several machines, and their location, storage replication, load balancing and functionality is not transparent. Distributed systems usually use some kind of client-server organization.” – FOLDOC “A Distributed System comprises several single components on different computers, which normally do not operate using shared memory and as a consequence communicate via the exchange of messages. The various components involved cooperate to achieve a common objective such as the performing of a business process.” – Schill & Springer Main characteristics – Components: Multiple spatially separated individual components Components posses own memory Cooperation towards a common objective – Resources: Access to common resources (e.g., databases, file systems) – Communication: Communication via messages – Infrastructure: Heterogeneous hardware infrastructure & software middleware 2

3. Distributed systems These definitions do not define the insides of a distributed system – Design and implementation – Maintenance – Algorithmics (i.e., protocols) Facebook social network graph among humans. The Internet color coded by ISPs. 3

4. A working definition “A distributed system (DS) is a collection of entities, each of which is autonomous, programmable, asynchronous and failure-prone, and which communicate through an unreliable communication medium.” Terms – Entity = process on a device (PC, server, tablet, smartphone) – Communication medium = wired or wireless network Course objective: – Design and implementation of distributed systems Source: https://courses.engr.illinois.edu/cs425/fa2013/lectures.html 4

5. The datacenter The datacenter lies at the foundation of many DSs – Amazon Web Services, Google Cloud, Microsoft Azure However, DSs can be comprised of PCs too. – P2P file sharing systems (e.g., Gnutella) Facebook’s Forest City Datacenter. 5

6. Example – Gnutella P2P What are the entities and communication medium? 6

7. Example – web domains What are the entities and communication medium? 7

8. The Internet Used by many distributed systems Vast collection of heterogeneous computer networks ISPs – companies that provide services for accessing and using the Internet Intranets – subnetworks operated by companies and organizations – Offer services unavailable to the public from the Internet – Can be the ISP’s core routers – Linked by backbones High bandwidth network links 8

9. Example - Intranet What are the entities and communication medium? 9

10. Parallel vs. distributed computing Parallelism – Perform multiple tasks at the same time – True parallelism requires distribution on multiple processors/cores/machines – Can range from many core to multi processor to many computer on shared or distributed memory Concurrency – Computations with multiple threads – Can exploit hardware parallelism but it is inherently related to the software need (i.e., react to different asynchronous events) – Concurrency becomes parallelism if parallelism is true (one thread per processor/core/machine) not virtual Distributed computing – Related to where the computation physically resides Distributed algorithm is executed on multiple CPUs, connected by networks, buses or any other data communication channel – Computers are connected by communication links on distributed memories Rely fundamentally on message passing – Usually part of the goal If resources are geographically spread than the system is inherently distributed 10

11. Parallel vs. distributed computing Is distributed computing a subset of parallel computing? Not an easy answer In favor – Distributed computing is parallel computing on geographically spread machines distributed  parallel  concurrent computing Against – They address different issues Distributed computing is focused on issues related to computation and data distribution Parallel computing does not address problems such as partial failures Parallel computing focuses on tightly coupled applications 11

12. Source: courses.washington.edu/css434/slides/w03w04/Fundamentals.ppt Parallel vs. distributed systems 12

13. Reasons for DS Inherently distributed applications – Distributed DB, worldwide airline reservation, banking system Information sharing among distributed users – CSCW or groupware Resource sharing – Sharing DB/expensive hardware and controlling remote lab. devices Better cost-performance ratio / Performance – Emergence of Gbit network and high-speed/cheap MPUs – Effective for coarse-grained or embarrassingly parallel applications MapReduce Reliability – Non-stopping (availability) and voting features. Scalability – Loosely coupled connection and hot plug-in Flexibility – Reconfigure the system to meet users ’ requirements 13

14. DS layered architecture Source: https://courses.engr.illinois.edu/cs425/fa2013/lectures.html Application e-mail remote terminal access Web file transfer streaming multimedia remote file server Internet telephony Application layer protocol smtp [RFC 821] telnet [RFC 854] http [RFC 2068] ftp [RFC 959] proprietary (e.g. RealNetworks) NFS proprietary (e.g., Skype) Underlying transport protocol TCP TCP or UDP typically UDP TCP=Transmission Control Protocol UDP=User Datagram Protocol Implemented via network “sockets”. Basic primitive that allows machines to send messages to each other Distributed System Protocols! Networking Protocols 14

15. Main issues of DS No global clock – No single global notion of the correct time (asynchrony) Unpredictable failures of components – Lack of response may be due to either failure of a network component, network path being down, or a computer crash (failure-prone, unreliable) Highly variable bandwidth – From 16Kbps (slow modems or Google Balloon) to Gbps (Internet2) to Tbps (in between DCs of same big company) Large and variable latency – Few ms to several seconds Large numbers of hosts – Up to several million Security and privacy – Due to geographical and political spread Interoperability – Due to various standards and protocols 15

16. DS design goals Heterogeneity – can the system handle a large variety of types of hardware and software (interoperability)? Robustness – is the system resilient to hardware and software crashes and failures, and to network dropping messages? Availability – are data & services always available to clients? Transparency – can the system hide its internal workings from users? Concurrency – can the server handle multiple clients simultaneously? Efficiency – is the service fast enough? Does it utilize 100% of all resources? Scalability – can it handle 100 million nodes without degrading service? (nodes=clients and/or servers) How about 6 B? More? Security – can the system withstand hacker attacks? Privacy – is the user data safely stored? Openness – is the system extensible? 16

17. History of distributed computing 1945-1950s Loading monitor 1950s-1960s Batch system 1960s Multiprogramming 1960s-1970s Time sharing systemsMultics, IBM360 1969-1973 WAN and LANARPAnet, Ethernet 1960s-early1980s MinicomputersPDP, VAX Early 1980s WorkstationsAlto 1980s – present Workstation/Server modelsSprite, V-system 1990s ClustersBeowulf Late 1990s Grid computingGlobus, Legion 2006 Cloud computingEC2 17

18. DS system models Minicomputer model Workstation model Workstation-server model Processor-pool model Cluster model Grid computing 18

19. Minicomputer Model Extension of time sharing system – User must log on his/her home minicomputer – Thereafter, he/she can log on a remote machine by telnet Resource sharing – Database – High-performance devices Mini- computer Mini- computer Mini- computer ARPA net 19

20. Workstation Model Process migration – Users first log on his/her personal workstation – If there are idle remote workstations, a heavy job may migrate to one of them Problems: – How to find am idle workstation – How to migrate a job – What if a user log on the remote machine 100Gbps LAN Workstation 20

21. Workstation-Server Model Client workstations – Diskless – Graphic/interactive applications processed in local – All file, print, http and even cycle computation requests are sent to servers Server minicomputers – Each minicomputer is dedicated to one or more different types of services Client-Server model of communication – RPC (Remote Procedure Call) – RMI (Remote Method Invocation) A client process calls a server process ’ function No process migration invoked Example: NSF 100Gbps LAN Workstation Mini- Computer file server Mini- Computer http server Mini- Computer cycle server 21

22. Processor-Pool Model Clients – They log in one of terminals (diskless workstations or X terminals) – All services are dispatched to servers Servers – Necessary number of processors are allocated to each user from the pool Better utilization but less interactivity Server 1 100Gbps LAN Server N 22

23. Cluster Model Client – Takes a client-server model Server – Consists of many PC/workstations connected to a high-speed network – Puts more focus on performance: Serves for requests in parallel 100Gbps LAN Workstation Master node Slave 1 Slave N Slave 2 1Gbps SAN http server1 http server2 http server N 23

24. High-speed Information high way Grid Computing Goal – Collect computing power of supercomputers and clusters sparsely located over the nation and make it available as if it were the electric grid Distributed supercomputing – Very large problems needing lots of CPU, memory, etc. High-Throughput computing – Harnessing many idle resources On-Demand computing – Remote resources integrated with local computation Data-intensive computing – Using distributed data Collaborative computing – Support communication among multiple parties Super- computer Cluster Super- computer Cluster Mini- computer Workstation 24

25. Internet Cloud Computing Goal – On demand virtualized access to hardware infrastructure – “pay per use” model for public clouds – “as a service” paradigm Several models – Infrastructure as a Service Clients manage virtualized resources – Amazon EC2, Google Cloud – Platform as a Service Clients have access to various platform services to develop, run, and manage applications without dealing with the infrastructure – Microsoft Azure – Software as a Service Clients have access only to specific software tools – GMail, Dropbox – Data as a Service Clients can access remotely stored data – Amazon Public Data Sets: sciences, economics. – … VM Database Specific services Workstation 25

26. What will you learn? Real distributed systems – Cloud computing Lectures 2 and 3 All labs (Amazon EC2, Google Cloud) – Hadoop MapReduce (lecture 2, labs 1 thru 4) – Key-value stores Lab 5 – Apache Storm Lecture 14 Labs 7 – P2P systems Lecture 10 Classical problems – Failure detection (lecture 4) – Time and synchronization (lecture 5) – Global states (lecture 6) – Multicast (lecture 7) – Leader election (lecture 9) – Networking and routing (lecture 11) – Gossiping (lecture 13) Concurrency – RPC (lecture 8 and lab 6) – Replication control (lecture 12) 26

27. Grading Oral evaluation (20%) – Questions from the lectures Projects (80%) – Assignments given during lab hours Documentation – Lecture slides, references inside the slides, Amazon AWS website, Hadoop website, Google. 27

28. Next lecture Introduction to cloud computing 28