I85PREP
⌘ K
Ruankao System Architect (Senior) study materials
Ruankao System Architect (Senior) study resources

Ruankao System Architect (Senior) practice questions

One question at a time. Build your Ruankao System Architect (Senior) understanding.

First 50 of 190 questions. Each one lists the choices, the correct answer, and an explanation. The checker at the top of this page saves progress in this browser when it is available.

Continue in the practice tool above to load more questions, 50 at a time.

Question 1Computer architecture, logic and data representation

A 软考 storage design needs one-disk-failure tolerance with block striping and distributed parity. Which RAID level matches that requirement?

  1. RAID 0
  2. RAID 1
  3. RAID 5 Correct answer.
  4. RAID 6

Explanation

RAID 5 stripes data and distributes parity across member disks, so the array can reconstruct data after a single disk failure. RAID 0 has no parity, RAID 1 mirrors without distributed parity, and RAID 6 uses dual parity to survive two failures.

Question 2Computer architecture, logic and data representation

Dijkstra’s banker’s algorithm is used in operating-system teaching to do which of the following?

  1. Detect a deadlock after a circular wait already exists
  2. Avoid deadlock by granting a request only when the resulting allocation remains safe Correct answer.
  3. Preempt CPU time slices to prevent starvation
  4. Replace pages using a least-recently-used stack

Explanation

The banker’s algorithm is a deadlock-avoidance method. Each process states a maximum claim; a request is granted only if there is still a sequence in which all processes can finish with the remaining resources (a safe state). Detection algorithms look for a cycle after deadlock exists; bankers do not schedule CPU or replace pages.

Question 3Computer architecture, logic and data representation

In classic semaphore teaching used by 软考, what does a P (wait) operation on semaphore S do when S is already 0?

  1. Increment S and continue immediately
  2. Decrement S below zero and continue without blocking
  3. Block the calling process until a later V operation makes a unit available Correct answer.
  4. Destroy the semaphore and wake every waiter

Explanation

P (wait/down) tries to acquire one unit. If S is 0, no unit is available, so the caller blocks until some other process executes V (signal/up) and increments S. P does not increment S, and it does not continue while S is 0.

Question 4Computer architecture, logic and data representation

In the TCP/IP model used on 软考 papers, TCP belongs to which layer?

  1. Network (internet) layer
  2. Transport layer Correct answer.
  3. Application layer
  4. Data-link layer

Explanation

TCP provides end-to-end reliable byte-stream service and is a transport-layer protocol. IP is the internet/network layer, HTTP/SMTP are application protocols, and Ethernet framing is data-link.

Question 5Computer architecture, logic and data representation

A campus LAN uses 192.168.10.0/26. How many usable host addresses are in one of these subnets?

  1. 62 Correct answer.
  2. 64
  3. 30
  4. 126

Explanation

A /26 prefix leaves 6 host bits, so 2^6 = 64 addresses per subnet. Subtract network and broadcast addresses to get 62 usable hosts. /27 would be 30 usable; /25 would be 126 usable; 64 counts every address including network and broadcast.

Question 6Computer architecture, logic and data representation

Belady’s anomaly, as tested in OS items, can appear with which page-replacement policy?

  1. Least recently used (LRU)
  2. Optimal (OPT/MIN)
  3. FIFO Correct answer.
  4. Clock with a referenced-bit approximation of LRU

Explanation

Belady’s anomaly is that more page frames can increase page faults. FIFO can exhibit it. Stack algorithms such as LRU and OPT never do, because the pages in a smaller memory are always a subset of those in a larger one.

Question 7Computer architecture, logic and data representation

A repairable service has MTTF = 99 hours and MTTR = 1 hour. Using the 软考 availability identity A = MTTF / (MTTF + MTTR), what is the steady-state availability?

  1. 0.99 Correct answer.
  2. 0.90
  3. 0.01
  4. 99

Explanation

Availability is the fraction of time the system is up: A = 99 / (99 + 1) = 99/100 = 0.99. 0.90 would be 90/(90+10); 0.01 is the downtime fraction; 99 is MTTF, not a dimensionless availability.

Question 8Computer architecture, logic and data representation

Two concurrent database transactions must not see each other’s uncommitted writes. Which ACID property is that requirement?

  1. Atomicity
  2. Consistency
  3. Isolation Correct answer.
  4. Durability

Explanation

Isolation (隔离性) means concurrent transactions behave as if serialized, so one does not read another’s dirty uncommitted data. Atomicity is all-or-nothing, consistency is invariant preservation, and durability is surviving a crash after commit.

Question 9Computer architecture, logic and data representation

A relation is in 2NF but a non-prime attribute depends on another non-prime attribute that itself depends on the primary key. Which normal form is violated?

  1. 1NF only
  2. 3NF Correct answer.
  3. BCNF is satisfied whenever 3NF is violated
  4. 5NF only

Explanation

Third normal form forbids transitive dependencies of non-prime attributes on the key. 1NF only concerns atomic values. BCNF is stricter than 3NF, so a 3NF violation is also a BCNF problem, not a sign that BCNF holds. 5NF concerns join dependencies, not this transitive case.

Question 10Computer architecture, logic and data representation

An architect must keep a log volume online after any two disks fail. Which RAID level is the standard match?

  1. RAID 0
  2. RAID 1 with two disks only
  3. RAID 5
  4. RAID 6 Correct answer.

Explanation

RAID 6 stores two independent parity sets, so any two disks can fail without data loss. RAID 0 has no redundancy. Two-disk RAID 1 survives only one disk failure. RAID 5 survives one failure, not two.

Question 11Computer architecture, logic and data representation

The classic TCP three-way handshake used to establish a connection is which sequence?

  1. ACK, then SYN, then FIN
  2. SYN, then SYN-ACK, then ACK Correct answer.
  3. FIN, then ACK, then SYN
  4. RST, then SYN, then ACK

Explanation

The client sends SYN, the server replies SYN-ACK, and the client completes with ACK. FIN is used to close; RST aborts; there is no ACK-first establishment sequence.

Question 12Operating systems, translation and concurrency

Which condition is NOT one of Coffman’s four necessary conditions for deadlock?

  1. Mutual exclusion
  2. Hold and wait
  3. Preemptable resources (resources can be forcibly taken) Correct answer.
  4. Circular wait

Explanation

Deadlock needs mutual exclusion, hold-and-wait, no preemption, and circular wait. If resources can be forcibly preempted, one of the four conditions fails and deadlock can be broken. The third Coffman condition is no preemption, not preemption.

Question 13Requirements, design, testing and architecture trade-offs

A project has stable, well-understood requirements and a contractual need to freeze scope before coding. Which process model is the best classic fit?

  1. Waterfall Correct answer.
  2. Extreme Programming with weekly story changes
  3. Throwaway prototyping as the delivery model
  4. Scrum with an open product backlog until release

Explanation

Waterfall proceeds through sequential phases and fits stable, specified requirements and contract-style freezes. XP and Scrum expect changing backlogs. Throwaway prototypes explore uncertainty; they are not the delivery model when scope is already frozen.

Question 14Requirements, design, testing and architecture trade-offs

Boehm’s spiral model is distinguished from a pure waterfall primarily by which idea?

  1. A single pass with no risk analysis
  2. Risk-driven iterations that combine prototyping with planned review of alternatives Correct answer.
  3. Eliminating documentation entirely
  4. Requiring pair programming as the only control

Explanation

The spiral model repeats a cycle of determine objectives, evaluate alternatives and risks, develop and verify, and plan the next iteration. Risk analysis and prototyping are central. It is not a single waterfall pass, and it does not ban documents or mandate XP practices.

Question 15Requirements, design, testing and architecture trade-offs

Which sequence is the four phases of the Rational Unified Process (RUP) in order?

  1. Construction, Transition, Inception, Elaboration
  2. Inception, Elaboration, Construction, Transition Correct answer.
  3. Elaboration, Inception, Transition, Construction
  4. Transition, Construction, Elaboration, Inception

Explanation

RUP’s four phases are Inception (scope and business case), Elaboration (architecture baseline and major risks), Construction (build the product), and Transition (deploy to users). Other orders scramble those goals.

Question 16Requirements, design, testing and architecture trade-offs

The Agile Manifesto values working software over which of the following?

  1. Individuals and interactions
  2. Customer collaboration
  3. Comprehensive documentation Correct answer.
  4. Responding to change

Explanation

The Manifesto values working software over comprehensive documentation (while still allowing documents). The other three phrases are themselves Manifesto left-hand values, not the things being de-emphasized relative to working software.

Question 17Requirements, design, testing and architecture trade-offs

In CMMI-DEV maturity levels used by 软考, which level first requires that processes are tailored from the organization’s standard process set?

  1. Level 1 Initial
  2. Level 2 Managed
  3. Level 3 Defined Correct answer.
  4. Level 5 Optimizing only, not Level 3

Explanation

Level 3 Defined is where the organization has a standard process set and projects tailor it. Level 1 is ad hoc. Level 2 Managed is project-level process discipline without an organization-wide standard set. Level 5 adds quantitative continuous improvement on top of Defined/Quantitatively Managed.

Question 18Requirements, design, testing and architecture trade-offs

GB/T 8566, as commonly paired with ISO/IEC 12207 in 软考 standards items, primarily specifies which of the following?

  1. Software life-cycle processes Correct answer.
  2. The Ethernet MAC frame layout
  3. The SQL:2016 grammar
  4. The Unicode character repertoire

Explanation

GB/T 8566 is the Chinese national standard for software life-cycle processes, aligned with ISO/IEC 12207. It does not define Ethernet frames, SQL syntax, or Unicode.

Question 19Requirements, design, testing and architecture trade-offs

In the V-model, system testing is planned to correspond most closely to which left-side activity?

  1. Coding of a single module
  2. System requirements / system design Correct answer.
  3. Daily stand-up meetings
  4. User-interface color selection only

Explanation

The V-model pairs each development activity with a test level. System testing validates the integrated system against system requirements (and related system design). Unit tests pair with module coding; stand-ups and color choices are not V-model pairings.

Question 20Requirements, design, testing and architecture trade-offs

A team recovers a design model from existing source without changing the running system. In 软考 software-maintenance vocabulary, that activity is called:

  1. Forward engineering
  2. Reverse engineering Correct answer.
  3. Business process reengineering of the whole enterprise
  4. Acceptance testing

Explanation

Reverse engineering extracts representations (design, architecture) from an existing implementation. Forward engineering builds an implementation from a model. BPR redesigns business processes, not source-to-model recovery. Acceptance testing checks the product against user criteria.

Question 21Requirements, design, testing and architecture trade-offs

“The payment API must respond within 200 ms at 5,000 requests per second.” In requirements classification used by architecture papers, this is primarily:

  1. A functional requirement describing a business action
  2. A non-functional (quality) requirement for performance Correct answer.
  3. A data-retention legal clause
  4. A user-interface branding guideline

Explanation

Response time and throughput are performance quality attributes, i.e. non-functional requirements. Functional requirements would say what the API does (charge a card). Retention and branding are different constraint classes.

Question 22Requirements, design, testing and architecture trade-offs

Stakeholders cannot articulate the UI workflow, and management forbids shipping prototype code. Which prototyping strategy is appropriate?

  1. Evolutionary prototyping that becomes the production system
  2. Throwaway (exploratory) prototyping discarded after requirements are understood Correct answer.
  3. No prototype; start waterfall coding of the unknown UI
  4. Big-bang cutover of an unvalidated full build

Explanation

Throwaway (exploratory) prototypes clarify unknown interaction requirements and are discarded so their shortcuts do not enter production. Evolutionary prototypes grow into the product, which management forbade. Skipping discovery and big-bang cutover both skip learning.

Question 23Requirements, design, testing and architecture trade-offs

In configuration management, a baseline is best described as:

  1. An informal personal working copy that anyone may overwrite
  2. A reviewed, frozen set of configuration items that later changes must go through change control Correct answer.
  3. A compiler warning level
  4. The product owner’s private backlog note

Explanation

A baseline is a formally identified, reviewed snapshot of configuration items. Subsequent changes are tracked through change control so the team can reconstruct what was agreed. Informal copies, compiler flags, and private notes are not CM baselines.

Question 24Requirements, design, testing and architecture trade-offs

CMMI Level 4 Quantitatively Managed is distinguished from Level 3 Defined primarily by which capability?

  1. Processes exist only in heroes’ heads
  2. Projects tailor an organization standard process, but without statistical process control
  3. Selected subprocesses are controlled using quantitative (statistical) techniques toward quality and performance objectives Correct answer.
  4. The organization abandons all defined processes

Explanation

Level 4 adds quantitative management: statistical understanding and control of selected subprocesses against numeric quality and process-performance objectives. Level 3 has a defined organizational process without that quantitative control. Level 1 is heroic; abandoning Defined processes would be a regression.

Question 25Middleware, distributed transactions and architecture evaluation

In 软考 middleware teaching, middleware is best placed where in the stack?

  1. Below the operating system kernel, replacing device drivers
  2. Between the operating system (and network) and distributed applications, providing reusable distribution services Correct answer.
  3. Inside a single spreadsheet cell formula
  4. As a replacement for TCP on the wire

Explanation

Middleware sits above the OS/network and below applications, offering location transparency, transactions, messaging, or component services. It does not replace the kernel or TCP, and it is not a spreadsheet formula.

Question 26Middleware, distributed transactions and architecture evaluation

Message-oriented middleware (MOM) is chosen primarily because it provides which property?

  1. Synchronous, tightly coupled RPC with a blocking caller
  2. Asynchronous, loosely coupled communication via queues or topics Correct answer.
  3. A single shared-memory address space for all processes on all hosts
  4. A replacement for disk storage

Explanation

MOM (for example JMS, MQ) decouples producers and consumers in time and space using queues or publish/subscribe. Callers need not block on the consumer. RPC is the opposite coupling style. MOM does not create a global shared memory or replace disks.

Question 27Middleware, distributed transactions and architecture evaluation

A bank core needs distributed two-phase commit across heterogeneous resource managers. Which middleware class is the classic 软考 match?

  1. Transaction-processing (TP) monitor / transaction middleware such as Tuxedo-style platforms Correct answer.
  2. A static HTML file server with no transaction API
  3. A pure CDN cache
  4. A compiler optimizer

Explanation

Transaction middleware / TP monitors coordinate distributed transactions (often XA/2PC) across resource managers. Static file servers, CDNs, and compilers do not provide that coordination.

Question 28Middleware, distributed transactions and architecture evaluation

In CORBA, the component that provides location transparency so a client invokes a remote object as if it were local is the:

  1. Object Request Broker (ORB) Correct answer.
  2. HTML renderer
  3. DNS root anycast only
  4. BIOS firmware

Explanation

The ORB marshals requests, finds the object implementation, and hides location and language details (with IDL). Browsers, DNS anycast, and BIOS are not the CORBA invocation bus.

Question 29Middleware, distributed transactions and architecture evaluation

In classic J2EE layering, which technology is intended for transactional business objects in the EJB container?

  1. Enterprise JavaBeans (EJB) Correct answer.
  2. Cascading Style Sheets only
  3. PNG image compression
  4. SMTP banner messages

Explanation

EJB components run in the EJB container and encapsulate transactional business logic. JSP/Servlets handle web presentation in the web container. CSS, PNG, and SMTP are not the J2EE business-object model.

Question 30Middleware, distributed transactions and architecture evaluation

Which statement matches 软考 application-server comparisons?

  1. Apache Tomcat is a servlet/JSP web container and is not a full Java EE EJB application server Correct answer.
  2. Tomcat includes a complete EJB container equivalent to WebLogic for every Java EE API
  3. WebLogic cannot run servlets
  4. A web container and an EJB container are the same process with identical contracts

Explanation

Tomcat implements Servlet/JSP (a web container). Full Java EE application servers such as WebLogic, WebSphere, or JBoss/WildFly add an EJB container and other Java EE services. Tomcat is therefore not a full Java EE stack.

Question 31Middleware, distributed transactions and architecture evaluation

An enterprise has dozens of point-to-point integrations that are expensive to change. Which platform is the 软考-style remedy for mediating services with routing and transformation?

  1. An Enterprise Service Bus (ESB) Correct answer.
  2. A single unversioned CSV file copied by hand
  3. A BIOS update
  4. Increasing the monitor refresh rate

Explanation

An ESB provides a mediated, often bus-topology integration layer (routing, transformation, protocol bridging) so applications do not each maintain N-to-N point-to-point links. CSV copies, BIOS, and display settings do not solve integration coupling.

Question 32Middleware, distributed transactions and architecture evaluation

A REST constraint that improves horizontal scaling of GET-heavy services is that each request:

  1. Stores conversation state only on the server session and never in the request
  2. Is stateless: all information needed to understand the request is in the request (and optional cache) Correct answer.
  3. Requires a sticky in-memory server session for every client forever
  4. Must use SOAP envelopes

Explanation

REST’s stateless constraint means the server does not rely on stored per-client conversation state to understand the request, which lets any replica handle it and enables caching. Sticky sessions and SOAP are not REST constraints.

Question 33Middleware, distributed transactions and architecture evaluation

Compared with REST as an architectural style, SOAP-based Web services are typically characterized by:

  1. XML envelopes, a WSDL contract, and a protocol that can run over HTTP (among other transports) Correct answer.
  2. A prohibition on using XML
  3. Being identical to the pipe-and-filter style
  4. Replacing DNS

Explanation

Classic WS-* stacks use SOAP XML messages and WSDL interface descriptions, commonly bound to HTTP. REST is a style with a uniform interface; SOAP is a protocol. SOAP uses XML rather than forbidding it, and it is not pipe-filter or a DNS replacement.

Question 34Middleware, distributed transactions and architecture evaluation

In JMS, which messaging domain delivers each message to at most one consumer from a queue?

  1. Publish/subscribe to a topic (one message, many subscribers)
  2. Point-to-point (queue) messaging Correct answer.
  3. Synchronous CORBA IIOP only
  4. SMTP carbon-copy

Explanation

JMS point-to-point uses a queue: one message is consumed by one receiver. Publish/subscribe on a topic fans out to many subscribers. CORBA IIOP and SMTP are different protocols.

Question 35Middleware, distributed transactions and architecture evaluation

A three-tier architecture (presentation, application/business, data) is chosen over two-tier fat-client C/S mainly to:

  1. Put business rules and connection pooling on a shared middle tier instead of duplicating them in every desktop client Correct answer.
  2. Guarantee that every client opens a raw database login with DBA privileges
  3. Eliminate the need for any network
  4. Force all logic into stored procedures with no application servers

Explanation

Three-tier moves business logic and resource management to a shared application tier, improving maintainability, pooling, and client thinness. It does not require DBA logins on every desktop, does not remove the network, and does not forbid an application server.

Question 36Middleware, distributed transactions and architecture evaluation

The classic SOA discovery/description/invocation trio taught with Web services is:

  1. UDDI, WSDL, and SOAP Correct answer.
  2. RAID, LRU, and RAID again
  3. SMTP, POP3, and IMAP only
  4. NTSC, PAL, and SECAM

Explanation

UDDI (directory), WSDL (service description), and SOAP (message protocol) are the textbook Web-services stack. RAID/LRU are storage/OS, mail protocols are mail, and NTSC/PAL are video standards.

Question 37Requirements, design, testing and architecture trade-offs

In 软考 architecture quality-attribute teaching (Bass-style), performance is primarily about:

  1. How long the system takes to respond, or how much work it completes in a time interval Correct answer.
  2. Whether unauthorized users are kept out
  3. How cheap a future change in functionality will be
  4. Whether a digital signature can be verified

Explanation

Performance is response time and/or throughput. Unauthorized access is security; cheap change is modifiability; signatures are a security/integrity mechanism.

Question 38Requirements, design, testing and architecture trade-offs

Availability, as a runtime quality attribute in 软考 texts, is best defined as:

  1. The proportion of time the system is up and able to deliver service, often related to MTTF and MTTR Correct answer.
  2. The number of features in the product backlog
  3. The number of concurrent users the licence permits
  4. The compiler vendor’s market share

Explanation

Availability is uptime proportion; tactics include redundancy, heartbeat, and restart. Feature count, manuals, and vendor share are not the availability attribute.

Question 39Requirements, design, testing and architecture trade-offs

A requirement says “adding a new payment channel must cost less than two person-weeks.” Which quality attribute is being specified?

  1. Modifiability Correct answer.
  2. Confidentiality only
  3. Throughput
  4. Mean time to hardware purchase

Explanation

Modifiability (可修改性) is the cost and risk of change. Confidentiality is a security sub-property; throughput is performance; hardware purchase time is not the software quality attribute.

Question 40Requirements, design, testing and architecture trade-offs

Bass-style security as a quality attribute is the ability to:

  1. Serve legitimate users while resisting unauthorized use, including confidentiality, integrity, and related controls Correct answer.
  2. Maximize cache hit ratio only
  3. Minimize the number of UML diagrams
  4. Guarantee that every algorithm is O(1)

Explanation

Security is providing service to authorized users while blocking unauthorized use; 软考 lists confidentiality, integrity, non-repudiation, and controllability among its facets. Cache hits are performance; UML count and O(1) are not the security attribute.

Question 41Requirements, design, testing and architecture trade-offs

In ATAM vocabulary used by 软考, a sensitivity point is:

  1. A property of one or more components (or their relationships) that is critical for achieving a particular quality attribute Correct answer.
  2. Any bug in a unit test
  3. Any defect found during system testing
  4. A legal disclaimer in the user license

Explanation

A sensitivity point is an architectural decision or component property that strongly affects one quality-attribute response. If that same property affects more than one attribute, it is also a tradeoff point. Bugs, slogans, and licenses are not ATAM sensitivity points.

Question 42Requirements, design, testing and architecture trade-offs

An architect increases encryption strength on every internal RPC. Latency rises while confidentiality improves. In ATAM terms this decision is primarily a:

  1. Tradeoff point (a sensitivity point for more than one quality attribute) Correct answer.
  2. Non-risk that cannot affect any quality attribute
  3. Page-replacement policy
  4. Compiler warning

Explanation

The same decision is sensitive for security (up) and performance (down), so it is a tradeoff point. It is not a non-risk, and it is unrelated to paging or compiler warnings.

Question 43Requirements, design, testing and architecture trade-offs

ATAM (Architecture Tradeoff Analysis Method) is used primarily to:

  1. Evaluate an architecture’s support for competing quality attributes and expose risks, sensitivity points, and tradeoffs before large implementation spend Correct answer.
  2. Replace all testing with a single unit-test file
  3. Generate machine code from UML automatically as its only purpose
  4. Set provincial 软考 fees

Explanation

SEI ATAM is a scenario-based architecture evaluation focused on quality-attribute tradeoffs. It does not replace testing, is not a code generator, and does not set exam fees.

Question 44Requirements, design, testing and architecture trade-offs

软考 teaching commonly groups ATAM into four phases. Which list is that grouping?

  1. Presentation; investigation and analysis; testing; reporting Correct answer.
  2. Coding; compiling; linking; shipping
  3. Inception, Elaboration, Construction, Transition only as ATAM phases
  4. Plan, Do, Check, Act as the ATAM phase names

Explanation

Tutorial presentations of ATAM used in 软考 list four phases: presentation (ATAM, business drivers, architecture), investigation and analysis (approaches, utility tree, analysis), testing (stakeholder scenarios), and reporting. RUP phases and PDCA are different frameworks; compile/link is a build pipeline.

Question 45Requirements, design, testing and architecture trade-offs

An ATAM utility tree is typically structured as:

  1. Root utility → quality attributes → refinements → quality-attribute scenarios as leaves, later prioritized Correct answer.
  2. A Gantt chart of vacation days
  3. A RAID configuration utility
  4. A compiler abstract-syntax tree of a single function

Explanation

The utility tree decomposes “utility” into quality attributes, then refinements, then scenarios that can be prioritized (often importance/difficulty). It is not a vacation Gantt, RAID tool, or compiler AST.

Question 46Requirements, design, testing and architecture trade-offs

In the layered architecture style, a layer should typically:

  1. Provide services to the layer above and consume services of the layer below (adjacent interaction) Correct answer.
  2. Call every other layer directly to minimize interfaces
  3. Share a single global mutable table with no access rules
  4. Disable all information hiding

Explanation

Layered style (a call-and-return family member) restricts interaction to adjacent layers, supporting modifiability and reuse. Bypass of layers, global mutable tables, and no hiding break the style’s constraints.

Question 47Requirements, design, testing and architecture trade-offs

In the pipe-and-filter style, filters should communicate primarily by:

  1. Incremental data streams through pipes, with each filter independent of its neighbors’ internals Correct answer.
  2. A single shared blackboard that every filter mutates arbitrarily
  3. Direct method calls into other filters’ private fields
  4. Broadcast of raw interrupts to the BIOS

Explanation

Pipes carry streams; filters transform input to output and should not know neighbors’ internals, which aids reuse and parallelism. Blackboards, private-field calls, and BIOS interrupts are other styles or non-architecture noise.

Question 48Requirements, design, testing and architecture trade-offs

How does batch-sequential processing differ from pipe-and-filter in Shaw/Garlan data-flow styles?

  1. Batch-sequential typically processes whole files/steps to completion before the next step; pipe-filter streams incremental data Correct answer.
  2. They are identical in every textbook
  3. Pipe-filter forbids any data transformation
  4. Batch-sequential requires an ORB

Explanation

Batch sequential (a data-flow substyle) waits for a complete batch at each step. Pipe-filter allows overlapping, incremental flow. They are not identical; filters do transform data; ORBs belong to distributed-object middleware.

Question 49Requirements, design, testing and architecture trade-offs

Event-driven (implicit invocation) style is chosen mainly because:

  1. Components announce events without naming the receivers, reducing direct coupling Correct answer.
  2. Every component must hard-code the name of every other component
  3. It forbids adding new listeners
  4. It requires a single-threaded BIOS

Explanation

Independent-component event systems register interest in events; announcers need not know subscribers, which supports extensibility. Hard-coding all names is tight coupling; the style exists to allow new listeners.

Question 50Requirements, design, testing and architecture trade-offs

A blackboard architecture is a repository substyle in which:

  1. Independent knowledge sources opportunistically update a shared data structure whose current state triggers the next contributions Correct answer.
  2. Each filter may only write to its private pipe
  3. Layers may only call the layer above
  4. Clients must use fat-client two-tier ODBC only

Explanation

Blackboard systems (classic AI/repository style) have a shared blackboard plus knowledge sources that activate based on the current state. Pipes, strict upward-only layering, and fat ODBC clients are other styles or technologies.