©NEC Laboratories America 1 Hui Zhang Samrat Ganguly Sudeept Bhatnagar Rauf Izmailov NEC Labs America Abhishek Sharma University of Southern California.

1©NEC Laboratories America

Hui Zhang Samrat Ganguly

Sudeept Bhatnagar Rauf Izmailov

NEC Labs America

Abhishek Sharma

University of Southern California

Optimal Load Balancing in Publish/Subscribe Broker Networksusing Active Workload Management


Outline

Problem statement Load balancing in pub/sub broker networks

Optimal load balancing half-cascading load distribution on a workload

aggregation tree

Shuffle Architecture workload balancing schemes

Analysis & Evaluation Conclusions


Publish/Subscribe Overlay Services

Publisher Y

Publisher X

Subscriber B

Subscriber A

Broker network

Subscription

Event


Workload Management In a Pub/Sub Broker Network

A broker network offers 1 function: message filtering. the process of selecting messages for reception.

4 types of workloads in a broker network. message parsing. message matching. message delivering. message forwarding.

Assumed the last to cause performance bottleneck.

1 unique factor in the difficulty of the workload management Run-time content matching

Our contribution: an active workload management middleware, offering optimal load balancing on all 4 types of the workloads. 2 main components: message shuffling and half-cascading

aggregation trees.


A Simple Optimal Load Balancing Scheme

A simple push-half-down load balancing scheme can be enabled with the workload aggregation tree as left.

An aggregation tree with the half-cascading load distribution under uniform traffic input distribution.


Message shuffling

Upon receiving a message m (event or subscription) from outside the broker network, the first assignment of a Shuffle node x is to redistribute it in the system. x will pick a random key for m (e.g., by hashing some

subscription ID contained in the message) and send it to the node y responsible for that key in the overlay space.

The above message shuffling achieves two goals: The randomization makes the distribution of the input traffic

for any potential aggregation tree uniform on the node space. Combing message shuffling, and Chord [stoica2001] with a new

node join/leave scheme, Shuffle can construct half-cascading aggregation trees.

The cost of message parsing on subscriptions is distributed evenly throughout the system so that Shuffle eliminates the potential performance bottleneck due to message parsing workload.


Shuffle – software architecture

The Shuffle node architecture


Shuffle – an example message filtering process

1. An event message e arrives from a publisher and on node x.

2. Node x forwards e to node y through message shuffling.

3. Node y parses e, and forwards the parsed message to each of the subscription aggregation trees that e’s attributes corresponds to.

4. In each aggregation tree, e is forwarded along the path from y to the root node following Chord routing protocol, and the node at each hop either forwards it or does message matching.

5. When the message matching is done, message delivering will be done in the same node afterwards.

6. Periodically, a load balancing process will be scheduled to balance the workload due to two independent inputs: streaming events and stored subscriptions.


Event Overload

0

1 2

3

X, Y0

1

23

X, Y/2

X, Y/2

X : # subs. for attribute AY: # of events with attribute A


Subscription Overload

X : # subs. for attribute A; Y: # of events with attribute A

b c

d

X,Y a

b c

d

X/2,Y

a

b c

d

X/4,Ya

b c

d

X/2,Y

X/4,YX/2,YX/4,Y

X/4,Y

X/4,Y

X/4,Y

a


Analysis

Result 1: when the Shuffle network size is a power of 2, every Shuffle node in any aggregation tree has the half-cascading load distribution on its children in terms of aggregated messages.

Result 2: When the Shuffle network size is not a power of 2, any non-leaf node x in an aggregation tree has at least one child which contributes no less than 1/4 of the total load aggregated on x.

Result 3: MIN-NODE-LOAD-FORWARD is NP-hard. MIN-NODE-LOAD-FORWARD: For a network of size N, given k

attribute trees, the number of subscriptions Xi at the root of each attribute tree i and threshold th, what is the minimum number of nodes in the network to which subscriptions must be transferred to such that the number of subscriptions at any node is at most th?


Evaluation

consider three load balancing schemes: Shuffle. Random-Half: In this scheme, an overloaded node picks

an underloaded node with random probing, and then splits half its load with that node. The overloaded node repeats the operation until its load is reduced below a target level.

Random-Min: Random-Min is the same as Random-Half except when an overloaded node splits its load with an underloaded node, it just delegates a bare minimum load equal to the target value to the chosen node by replicating its subscription set there and forwarding a commensurate fraction of event traffic there.


Single aggregation tree results (1)

Event load balancing – Control Messages


Single Aggregation Trees Results (2)

Event load-balancing- Message Forwarding Load


Multiple Aggregation Tree Results

Subscription load balancing - Nodes affected


Conclusions

In this paper, we present the design of Shuffle, an active workload management middleware to support a scalable broker network. Shuffle offers an integral solution to manage all types of the

workload in a pub/sub broker network. The load balancing performance is insensitive to the data

distribution of input requests. The load balancing does not introduce extra maintenance

cost on the overlay topology.

©NEC Laboratories America

Thank you!

Questions?


Backup - 1

©NEC Laboratories America 1 Hui Zhang Samrat Ganguly Sudeept Bhatnagar Rauf Izmailov NEC Labs America Abhishek Sharma University of Southern California.

Documents

message matching

parsed message

message forwarding

message parsing workload

event message e

nec laboratories america

combing message shuffling

cost of message parsing