Dutch Bangla Bank; Probationary Officer (Networking) Written Test: 2017প্রশ্ন ও সমাধান

DoS (Denial of Service): an attack, launched from a single source, that overwhelms a target's resources (bandwidth, CPU, memory, connection table) or exploits a software flaw, so legitimate users can no longer access the service. Examples: SYN flood, Ping of Death, Slowloris.

DDoS (Distributed Denial of Service): the same goal, but the flood is launched simultaneously from many distributed, compromised machines (a botnet) controlled by the attacker, so the traffic is far larger and its sources are spread across many IPs/networks — much harder to block by simply banning one address.

graph LR
    A[Attacker] --> C2[Command & Control Server]
    C2 --> Z1[Bot 1]
    C2 --> Z2[Bot 2]
    C2 --> Z3[Bot N]
    Z1 --> T[(Target Server)]
    Z2 --> T
    Z3 --> T

Aspect	DoS	DDoS
Source	One attacking machine	Many compromised machines (botnet)
Traffic scale	Limited by one host's capacity	Very large, aggregated across bots
Blocking	Easy — block one IP	Hard — thousands of distinct IPs
Typical technique	SYN flood, Ping of Death	UDP/DNS amplification, Smurf, botnet HTTP flood

Mitigation: rate limiting, firewalls/IPS rules, traffic-scrubbing services, CDNs/anycast to absorb and distribute load, and SYN cookies against connection-exhaustion attacks.

Exam takeaway: DDoS is DoS "at scale" via a botnet — defenses shift from simple IP blocking to traffic-pattern-based filtering.

Big Data refers to datasets so large, fast-growing, or structurally complex that traditional relational databases and single-machine tools cannot efficiently capture, store, manage, or analyze them — requiring distributed storage and parallel-processing frameworks instead.

The 5 V's that characterize it:

• Volume — sheer scale of data (terabytes to petabytes and beyond), e.g. social media posts, IoT sensor streams.
• Velocity — speed at which data is generated and must be processed, often in real time.
• Variety — structured (tables), semi-structured (JSON, XML), and unstructured (images, video, free text) data mixed together.
• Veracity — uncertainty, noise, and trustworthiness of the data.
• Value — the actual business/analytical insight that makes collecting all this worthwhile.

Example: a telecom operator analyzing billions of call-detail records per day from millions of subscribers, in near real time, to detect fraud — no single machine or ordinary RDBMS can ingest and query that volume/velocity, so it is ingested via a message queue and processed on a distributed cluster.

graph LR
    DS["Data Sources: IoT, Social, Logs"] --> ING["Ingestion: Kafka / Flume"]
    ING --> ST["Storage: HDFS / Data Lake"]
    ST --> PR["Processing: Spark / MapReduce"]
    PR --> AN["Analytics / BI / ML"]

Common tools: Hadoop (HDFS + MapReduce), Apache Spark, NoSQL stores (MongoDB, Cassandra), and Kafka for streaming ingestion.

Exam takeaway: Big Data is defined by the 5 V's, not size alone — the standard solution is distributed storage plus parallel processing, not scaling up a single relational database.