Information Technology: Wireless, internet and cloud

The answer is the second option. 'Serverless' does not mean there are no servers; it means the developer does not manage them. In Function as a Service (FaaS), a developer uploads a piece of code, a function, and specifies the events that should trigger it: a file arriving in storage, a message from an IoT device, a web request or a timer. The provider allocates computing capacity when the event occurs, scales from zero to thousands of parallel copies if needed, and bills by the number of invocations and the time they actually run, so nothing is paid while the code is idle. This makes serverless popular for bursty workloads such as processing IoT sensor messages, resizing uploaded images or running scheduled jobs. The provider also patches the underlying OS and runtime, while the developer remains responsible for the code, its permissions and the data it handles.

The answer is processing near the source. In a pure cloud design, every reading travels to a distant data centre, is processed there and a decision travels back. For a self- driving vehicle, a factory robot or a grid protection relay, that round trip can be too slow, and streaming raw video or high-frequency vibration data to the cloud wastes bandwidth. Edge computing moves part of the computation to the device, a gateway or a nearby micro data centre, so that time-critical decisions are taken locally with very low latency, only summaries or alerts go to the cloud, and the system keeps working if the link fails. Processing locally can also help privacy, since raw images or personal data need not leave the premises. The cloud is still used for heavy analytics, long-term storage and training models that are then pushed to the edge.

Statements 1 and 2 are correct. NIST SP 800-145 lists four deployment models: private, community, public and hybrid. A community cloud serves a group of organisations with shared concerns (mission, security, policy or compliance) and may be owned by one or more of them, a third party or a combination; a government cloud shared by several departments is a typical example. A hybrid cloud joins two or more distinct infrastructures that remain separate entities but are bound by standardised or proprietary technology enabling data and application portability, for example 'cloud bursting' from a private cloud into a public one at peak load. Statement 3 is wrong. 'Public' describes who may use the cloud (the general public), not who owns it; NIST notes that a public cloud may be owned and operated by a business, academic or government organisation, or a combination, and in practice most public clouds are run by private companies.

Statement (d) is the incorrect one. The first three are three of the five essential characteristics in NIST SP 800-145: rapid elasticity, measured service and broad network access (the other two are on-demand self-service and resource pooling). Statement (d) ignores the shared responsibility model. Providers secure the infrastructure they run, but customers always keep responsibility for their own data, user accounts and access permissions, and, depending on the service model, also for operating systems, applications and network rules. Industry reports and CERT-In advisories repeatedly trace cloud data leaks to misconfiguration by customers, such as publicly exposed storage buckets or weak access keys, rather than to failures of the provider's hardware.

Only items 1 and 2 are essential characteristics. NIST SP 800-145 (September 2011) lists five: on-demand self-service (a user can provision computing capability without human interaction with the provider), broad network access, resource pooling, rapid elasticity (capacity scales out and in, often automatically, so it appears unlimited) and measured service (use is metered, which makes pay-per-use possible). Item 3 is the opposite of cloud: the consumer rents capability and does not own the provider's servers. Item 4 contradicts resource pooling, under which the provider's resources serve many consumers through a multi-tenant model, with physical and virtual resources assigned and reassigned according to demand. Dedicated hardware can be bought as an option, but it is not a defining characteristic.

Statements 1 and 2 are correct. GI Cloud, named MeghRaj, was launched by the then Department of Electronics and IT on 4 February 2014 so that ministries could share computing infrastructure, deploy applications faster and avoid building separate data centres for every project. To give departments a trusted list of commercial providers, MeitY began empanelling the cloud service offerings of providers from 2016; a provider must meet the pre- qualification and compliance requirements and its data centres and offerings are audited by STQC (Standardisation Testing and Quality Certification), with surveillance audits afterwards. Statement 3 is wrong: the empanelment system exists precisely so that departments can procure services from empanelled private providers as well as from NIC's national cloud, subject to MeitY's security and data-location conditions.

Statements 1 and 2 are correct. A hypervisor creates and runs virtual machines (VMs), each with its own guest operating system. A Type 1 or bare-metal hypervisor sits directly on the hardware and is the usual choice in cloud data centres; a Type 2 or hosted hypervisor runs as an application on a desktop operating system, as in many personal virtualisation tools. NIST SP 800-190 explains that containers package an application with its libraries but share the host OS kernel, isolated from one another by operating-system features. Because no separate guest OS has to boot, containers are small and start in seconds, which suits microservices and auto-scaling. Statement 3 reverses the trade-off: precisely because they share one kernel, containers have weaker isolation than VMs, whose hypervisor enforces hardware-level separation; a kernel flaw can affect all containers on the host. That is why sensitive workloads often run containers inside VMs.

Statements 1 and 2 are correct. Photon energy is E = hf, so it rises with frequency. Visible light is around 4-7.5 x 10^14 Hz, about ten thousand times higher than 26 GHz (2.6 x 10^10 Hz), so a visible photon carries far more energy. Ionisation needs photon energies of roughly 10 eV and above, reached only in the far ultraviolet, X-rays and gamma rays. All mobile bands, mmWave and even terahertz, lie far below, which is why they are classed as non- ionising; their main biological effect at high intensity is heating.