Lecture_2_Architectures of Distributed System.pptx
- 1. /26 1 Parallel and Distributed Computing Lecture 2: Architectures of Distributed System Zubaer Ibna Mannan Kyungdong University Global Campus
- 3. /26 3 Architectural Styles Architectural style is formulated in terms of components. A component is a modular unit with well-defined required and provided interfaces that is replaceable within its environment. The way that components are connected to each other, the data exchanged between components, and finally how these elements are jointly configured into a system. In a distributed system a connector is generally described as a mechanism that mediates communication, coordination, or cooperation among components Based on the components and connectors, the architectural styles of a distributed system can be classified as Layered architectures Object-based architectures Data-centered architectures Event-based architectures
- 4. /26 4 Layered Architectures Components are organized in a layered fashion. A component at layer LN is allowed to call components at the underlying layer LN −1, but not the other way around. A key observation is that control generally flows from layer to layer. Requests go down the hierarchy In contrast, the results flow upward.
- 5. /26 5 Object-based Architectures Objected-based architectural styles is based on an arrangement of loosely coupled objects, it is less structured. Each object corresponds to what we define as a component. These components are connected through a (remote) procedure call mechanism. In this case of distributed systems, the calling object does not need to be executed on the same machine as the called object. Object-based architectures are attractive because they provide a natural way to encapsulate data (called the “state” of the object) and the operations that can be performed on that data (which are called “methods” or behavior of the object) in a single entity.
- 6. /26 6 Data-centered Architectures In data-centered architecture, the data is centralized and accessed frequently by other components, which modify data. The main purpose of this style is to achieve integrality of data. Data-centered architecture consists of different components that communicate through shared data repositories. The components access a shared data structure and are relatively independent, in that, they interact only through the data store. The most well-known examples of the data-centered architecture is a database architecture, in which the common database schema is created with data definition protocol.
- 7. /26 7 Event-based Architectures In event-based architectures, processes essentially communicate through the propagation of events and optionally also carry data. For distributed systems, event propagation has generally been associated with publish/subscribe systems. The basic idea is that processes publish events after which the middleware ensures that only those processes that subscribed to those events will receive them. The main advantage of event-based systems is that processes are loosely coupled. In principle, they need not explicitly refer to each other. Event-based architectures can be combined with data- centered architectures, yielding as shared data spaces.
- 9. /26 9 Centralized Architectures: Client – Server Client – Server model processes in a distributed system are divided into two (possibly overlapping) groups. A server is a process implementing a specific service, for example, a file system service or a database service. A client is a process that requests a service from a server by sending it a request and subsequently waiting for the server’s reply. When a client requests a service, it simply packages a message for the server, identifying the service it wants, along with the necessary input data. The message is then sent to the server. The latter, in turn, will always wait for an incoming request, subsequently process it, and package the results in a reply message that is then sent to the client. Have multiple clients that decide when and how to use the shared resource and display it, change data, and send it back to the server.
- 10. /26 10 Three-tier architecture organizes applications into three logical and physical computing tiers: the presentation tier, or user interface; the application tier, where data is processed; and the data tier, where the data associated with the application is stored and managed. Centralized Architectures: Three – tier Presentation tier is the user interface and communication layer of the application, where the end user interacts with the application. Its main purpose is to display information to and collect information from the user. Application tier (logic tier or middle tier) is the heart of the application. Information collected in the presentation tier is processed and decision making is conducted in middle tier. It act like an agent that receives requests from clients, processes the data and then forwards it on to the servers. Data tier (or database tier) the information processed by the application is stored and managed. In a three-tier application, all communication goes through the application tier. The presentation tier and the data tier cannot communicate directly with one another.
- 11. /26 11 Centralized Architectures: Multi – Tier Multi – Tier architecture is used to divide an enterprise application into two or more components that may be separately developed and executed. Multi-tier application includes: presentation tier, application processing tier, data access tier, and data tier. Presentation tier: Provides basic user interface and application access services Application processing tier: Possesses the core business or application logic Data access tier: Provides the mechanism used to access and process data Data tier: Holds and manages data that is at rest This division allows each component/tier to be separately developed, tested, executed and reused. Enterprise application has both external and internal users which use a web- based application which communicates back to a common set of application servers. Finally, the application servers communicate with a database server.
- 12. /26 12 In Peer-to-Peer (P2P) architectures, computing or networking is a distributed application architecture that partitions tasks or workloads between peers. Peers are equally privileged, equipotent participants in the application and said to form a peer-to-peer network of nodes. This means that the functions that need to be carried out are represented by every process that constitutes the distributed system. De-centralized Architectures: Peer-to-Peer (P2P) Peers make a portion of their resources (such as processing power, disk storage or network bandwidth) directly available to other network participants, without the need for central coordination by servers or stable hosts. Peers are both suppliers and consumers of resources, in contrast to the traditional client–server model in which the consumption and supply of resources is divided. Peer
- 13. /26 13 De-centralized Architectures: Unstructured peer-to-peer Unstructured peer-to-peer networks do not impose a particular structure on the overlay network by design, but rather are formed by nodes that randomly and freely connects with each other, e.g., it selects arbitrary peers as neighbors. It neither has a centralized directory nor has any control over the network topology or resource placement. It largely relies on randomized algorithms for constructing an overlay network. Like as randomized neighboring nodes the data items are assumed to be randomly placed on these nodes. Live node Each node maintains a list of c neighbors (also known partial view). Nodes regularly exchange entries from their partial view. Each entry identifies another node in the network, and has an associated age that indicates how old the reference to that node is. The active thread takes the initiative to communicate with another node and selects the nodes from its current partial view. Assuming that entries need to be pushed to the selected peer, it continues by constructing a buffer containing c/2+1 entries, including an entry identifying itself. The other entries are taken from the current partial view.
- 14. /26 14 De-centralized Architectures: Unstructured peer-to-peer If the node is also in pull mode it will wait for a response from the selected peer. That peer, in the meantime, will also have constructed a buffer by means the passive thread whose activities strongly resemble that of the active thread. Benefits: Because of no structure globally imposed upon them, unstructured networks are easy to build and allow for localized optimizations to different regions of the overlay. Also, because of same role of all peers in network, unstructured networks are highly robust. Live node Limitations: The primary limitations of unstructured networks also arise from this lack of structure. When a peer wants to find a desired piece of data in the network, the search query must be flooded through the network to find as many peers as possible that share the data. Flooding causes a very high amount of signaling traffic in the network, uses more CPU/memory and does not ensure that search queries will always be resolved. Furthermore, since there is no correlation between a peers and the content managed by it, there is no guarantee that flooding will find a peer that has the desired data.
- 15. /26 15 In Structured Peer-to-Peer architectures, the overlay is organized into a specific topology, and the protocol ensures that any node can efficiently search the network for a file/resource, even if the resource is extremely rare. The overlay network is constructed using a deterministic procedure, namely distributed hash table (DHT). In DHT, a variant of consistent hashing is used to assign ownership of each file to a particular peer and enables peers to search for resources on the network using a hash table: that is, (key, value) pairs are stored in the DHT. De-centralized Architectures: Structured Peer-to-Peer In order to route traffic efficiently through the network, nodes in a structured overlay must maintain lists of neighbors that satisfy specific criteria. This makes them less robust in networks with a high rate of churn (i.e. with large numbers of nodes frequently joining and leaving the network) When looking up a data item, the network address of the node responsible for that data item is returned. This is accomplished by routing a request for a data item to the responsible node.
- 16. /26 16 Hybrid Architectures (1) Many distributed systems combine architectural features (example: client-server solutions are combined with decentralized architectures), such combine architectural featuring systems are known as Hybrid architectures. Edge-Server Systems: are deployed on the Internet where servers are placed ‘‘at the edge’’ of the network. This edge is formed by the boundary between enterprise networks and the actual Internet (example. Internet service provider (ISP)). End users at home connect to the Internet through their ISP. End users, or clients in general, connect to the Internet by means of an edge server. The edge server’s main purpose is to serve content, possibly after applying filtering and transcoding. More interesting fact is that a collection of edge servers can be used to optimize content and application distribution. For a specific organization, one edge server acts as an origin server from which all content originates. That server can use other edge servers for replicating Web pages.
- 17. /26 17 Hybrid Architectures (2) Collaborative Distributed Systems is used to deploy the Hybrid structures. Main issue of Hybrid systems is to get started first which often use centralized client-server scheme. Once a node (client-server) has joined the system, it uses a fully decentralized scheme for collaboration. Example: BitTorrent file-sharing. BitTorrent is a peer-to-peer file downloading system where an end user downloads chunks of the desired file from other users until the downloaded chunks can be assembled together yielding the complete file. In most file-sharing systems, a file can be downloaded only when the downloading client is providing content to someone else. To download a file, a user needs to access a global directory (i.e., Web sites) where such a directory contains references (.torrent files). A .torrent file contains the information that is needed to download a specific file. Particularly, it refers to a tracker which is a server that keeps an accurate account of active nodes that have chunks of the requested file. An active node is one that is currently downloading another file. Once the nodes have been identified from where chunks can be downloaded, the downloading node effectively becomes active and forces other nodes to help. If node P notices Q is downloading more than uploading, then P can decrease the rate at which P sends data to Q.
- 19. /26 19 Middleware System What Does Middleware Mean? Middleware is a software layer situated between applications and operating systems. Middleware is typically used in distributed systems where it simplifies software development by doing the following: Hides the intricacies of distributed applications Hides the heterogeneity of hardware, operating systems and protocols Provides uniform and high-level interfaces used to make interoperable, reusable and portable applications Provides a set of common services that minimizes duplication of efforts and enhances collaboration between applications
- 20. /26 20 Kinds Middleware System (1) New application development Middleware can support modern and popular runtimes for a variety of use cases. Developers and architects can work with agility across platforms, following sets of foundational runtimes, frameworks, and programming languages. Middleware can also deliver commonly used functions such as web servers, single sign-on (SSO), messaging, and in-memory caching. Optimization of existing applications Middleware can help developers transform legacy monolithic applications into cloud-native applications, keeping valuable tools active with better performance and more portability. Legacy applications are implemented as monoliths, in which all data storage and processing are controlled by the monolith, and all functions for all data objects are processed using the same backend codebase. Monolith
- 21. /26 21 Kinds Middleware System (2) Comprehensive integration Middleware integration tools connect critical internal and external systems. Integration capabilities like transformation, connectivity, composability, and enterprise messaging, combined with SSO authentication, make it easier for developers to extend capabilities across different applications. Application Programming Interfaces (APIs) Many middleware services are accessed through APIs, which are sets of tools, definitions, and protocols that allow applications to communicate with each other. APIs make it possible to connect completely different products and services through a common layer.
- 22. /26 22 Kinds Middleware System (3) Data streaming While APIs are one way to share data between applications, another approach is asynchronous data streaming. This replicates a data set in an intermediate store, where the data can be shared among multiple applications. One popular open source middleware tool for real-time data streaming is Apache Kafka. Intelligent business automation Middleware can help developers, architects, IT, and business leaders automate manual decisions. Automation can improve resource management and overall efficiency.
- 23. /26 23 SELF-MANAGEMENT IN DISTRIBUTED SYSTEMS
- 24. /26 24 The Feedback Control Model The most important feature of self-managing systems is the adaptations which take place by means of one or more feedback control loops. The core of a feedback control system is formed by the components that need to be managed. These components are assumed to be driven through controllable input parameters, but their behavior may be influenced by all kinds of uncontrollable input, also known as disturbance or noise input. Although disturbance will often come from the environment in which a distributed system is executing, it may well be the case that unanticipated component interaction causes unexpected behavior. There are essentially three elements that form the feedback control loop: Metric Estimation Component Feedback Analysis Component Adjustment Component
- 25. /26 25 The Feedback Control Model Metric Estimation Component: Measures the various aspects (such as round-trip delays, latency between nodes, effect of noise, placing replicas, changing scheduling priorities, switching services, and so on) of the distributed system and sends those measured metrics to the analyzer. Feedback Analysis Component: It analyzes the measurements and compares these to reference values. It forms the heart of the control loop as it contain the algorithms that decide on possible adaptations. Adjustment Component: consist of various mechanisms to directly influence the behavior of the system. There can be many different mechanisms: placing replicas, changing scheduling priorities, switching services, moving data for reasons of availability, redirecting requests to different servers, etc. The analysis component will need to be aware of these mechanisms and their expected effect on system behavior. Therefore, it will trigger one or several mechanisms, to subsequently later observe the effect.
- 26. /26 26 Home Work Study Examples: 2.4.2 Example: Systems Monitoring with Astrolabe 2.4.3 Example: Differentiating Replication Strategies in Globule 2.4.4 Example: Automatic Component Repair Management in Jade