Reliable Communication

Smita HiremathCSC 8530

Reliable Client-Server Communication

Point-to-Point communication Established by TCP

Masks omission failure, crash failure not masked

Cont…

RPC Semantic in the presence of Failure1. The client is unable to locate the server2. The request message from the client to

the server is lost3. The server crashes after receiving a

request4. The reply message from the server to the

client is lost5. The client crashes after sending the

request.

Cont…

Client Cannot Locate the Server Server might be down Server evolves and new version of interface is

installed

Lost Request Messages Easiest one to deal Client stub start the timer when sending the

request.

Cont…

Server Crashes

A server in client-server communicationa) Normal caseb) Crash after execution c) Crash before execution

Cont…

At least once semantics – Guarantees that the RPC has been carried out at least once, but possibly more.

At most once semantics – Guarantees that the RPC has been carried out at most once, but possibly none at all

Guarantees nothing. RPC may have been carried out anywhere from zero to a large number of times

Cont…

Lost Reply Message Some requests can be repeated (idempotent) Assign each request a sequence number

Client Crashes Computations are orphans Extermination – the log is kept on the disk or some

other medium that survives crash. Reincarnation – divide the time up into sequential

numbered epochs.

Cont… Gentle reincarnation – computations are killed if the

owner cannot be found. Expiration – RPC is given standard amount of time.

Reliable Group Communication

Basic Reliable-Multicasting Message sent to a process group should be

delivered to each member of that group. Agreement exists on who is the a member of

the group.

Reliable-Multicasting scheme

A simple solution to reliable multicasting when all receivers are known and are assumed not to fail

a) Message transmissionb) Reporting feedback

Scalability in Reliable Multicasting

Two different approaches Nonhierarchical Feedback Control Hierarchical Feedback Control

Nonhierarchical Feedback Control

Scalable Reliable Multicasting (SRM) Never acknowledges successful delivery of

multicast messages. If the receiver missed a message, it multicasts

its feedback to the rest of the group, which allows another group member to suppress its own feedback.

Only a single feedback message will reach the sender, which in turn subsequently retransmits the message.

Nonhierarchical Feedback Control

Several receivers have scheduled a request for retransmission, but the first retransmission request leads to the suppression of others.

Scale reasonably well Problems with this approach

Ensuring that only one request for retransmission is returned to the sender requires accurate scheduling of feedback messages at each receiver.

Feedback also interrupts those processes to which the message has been successfully delivered.

Nonhierarchical Feedback Control

Hierarchical Feedback Control Scalability for very large groups of receivers. The group of receivers are partitioned into

number of subgroups, which are organized into trees.

Each subgroup appoints a local coordinator, which is responsible for handling retransmission

Local coordinator will have its own history buffer.

The essence of hierarchical reliable multicasting.a) Each local coordinator forwards the message

to its children.b) A local coordinator handles retransmission

requests.

Hierarchical Feedback Control

Atomic Multicast

Atomic multicast problem – message must be delivered to either all processes or none at all.

Virtual Synchrony

The logical organization of a distributed system to distinguish between message receipt and message delivery

Cont…

The principle of virtual synchronous multicast.

Message Ordering

Four different orderings: 1. Unordered multicasts2. FIFO-ordered multicasts3. Causally-ordered multicasts4. Totally-ordered multicasts

Cont…

Process P1 Process P2 Process P3

sends m1 receives m1 receives m2

sends m2 receives m2 receives m1

Three communicating processes in the same group. The ordering of events per process is shown along the vertical axis.

Cont…

Process P1 Process P2 Process P3 Process P4

sends m1 receives m1 receives m3 sends m3

sends m2 receives m3 receives m1 sends m4

receives m2 receives m2

receives m4 receives m4

Four processes in the same group with two different senders, and a possible delivery order of messages under FIFO-ordered multicasting

Cont…Multicast

Basic Message Ordering

Total-ordered Delivery?

Reliable multicast None No

FIFO multicastFIFO-ordered delivery

No

Causal multicastCausal-ordered delivery

No

Atomic multicast None Yes

FIFO atomic multicast

FIFO-ordered delivery

Yes

Causal atomic multicast

Causal-ordered delivery

Yes

Six different versions of virtually synchronous reliable multicasting.

Implementing Virtual Synchrony

a) Process 4 notices that process 7 has crashed, sends a view changeb) Process 6 sends out all its unstable messages, followed by a flush

messagec) Process 6 installs the new view when it has received a flush message from

everyone else