Summary questions on distributed systems¶

Questions are randomized per student, and you will only receive correct/incorrect feedback for them. In Problem 1, each question has exactly one correct option.

In Problem 2, each student will get at least one correct option, usually more. Problem 2 is accepted when the score is 12 or higher, and points are shown with a yellow background. Unfortunately, Plussa gives “incorrect” feedback for partially correct answers. However, Problem 2 is considered passed if the score is 12 or more. From the score alone, you cannot directly infer whether an individual choice was correct or not. Points start accumulating only after more than half of the choices in the problem are correct.

Question 1

Where is a distance function needed in distributed systems?

To find a neighbour in a structured P2P system.

To choose a neighbour in an unstructured P2P system.

To identify an attack target based on a changed distance.

To create a wormhole as a shortcut in the overlay.

Question 2

Not so much for information dissemination but for information retrieval, the most efficient type of distributed system is

a system that coordinates resources.

a system that coordinates services.

structured P2P.

unstructured P2P.

Question 3

A distributed hash table, DHT,

is the core of all P2P networks.

is also suitable in coordinated systems for storing addresses.

is the most important in structured P2P networks.

is suitable outside distributed systems for storing password hashes.

Question 4

Although blockchains belong to the advanced part of the material, everyone should know that their essential structure is a hash chain, i.e. the block’s hash digest is part of the next block. How does such a structure compare to a distributed hash table (DHT)?

They have nothing to do with each other, since the “hash” in DHT means distribution and not a hash digest.

In both cases the hash is a similar cryptographic function, but otherwise the structures are different.

The elements of a DHT also form a chain in which each element is a hash value of another.

They are two different names for the same thing in different application areas.

Question 5

A hierarchical P2P protocol

is not P2P at all, because nodes are not in a peer‑to‑peer position relative to each other.

is both P2P and a resource‑coordinated distributed system at the same time.

is another name for a hybrid P2P model that combines unstructured and structured features.

puts some more capable nodes in a special role but is still a peer network.

Question 6

Which of the following parts of the attack surface is the most characteristic of distributed systems?

access control supported by naming, identification, and authentication.

data transfer between distributed resources.

the resource coordination scheme.

data management that supports distributed applications.

Question 7

An Eclipse attack

is a special case of a Sybil attack.

is a concept that includes a Sybil attack as a special case.

must be carried out before a Sybil attack is possible.

may be possible by means of a Sybil attack.

Question 8

The whole module on distributed systems is fairly abstract (including the advanced parts). What is the reason for this? (The question does not assume you knew in advance, but you should understand the answer choices—and not believe all of them. Nor do you need to read the advanced parts for this.)

There are many kinds of distributed systems and many do not belong to basic ICT competence, so a comprehensive background for security properties is abstract.

An abstract treatment fits because the main focus is security anyway, which is presented widely with very concrete examples.

Resource distribution belongs in practice to operating systems and virtualization, where this material also covers it.

Distributed systems are mostly either historical, like Napster and many other peer networks, or still theoretical, like Chord and CAN.

Select the correct statements. You may select one or more options.

Question 1

P2P resembles more consumer-to-consumer trade than business-to-business trade.

P2P is almost identical in nature to a multi-tier client-server architecture.

Unstructured P2P is roughly the same as a peer network that operates without a protocol.

A structured P2P network forms its own address space.

The concept of overlay network applies similarly to both coordinated and uncoordinated distributed systems.

Blockchains are anonymous structured P2P networks.

A system that appears centralized is the goal when building a distributed system based on a coordination scheme.

A reputation system is an important access-control mechanism in a coordinated distributed system.

If admission control and access control are distinguished within the overall access-control model, a brand-new user of the system encounters them in exactly that order.

In distributed systems, in addition to side channels, there can be covert channels that break confidentiality.

There are only two kinds of distribution mechanisms related to databases: a ledger implemented with a blockchain and storage replication to RAID disks provided by the operating system.