Transcript Using Heterogeneous Paths for Inter

Using Heterogeneous Paths for Inter-process
Communication in a Distributed System
Vimi Puthen Veetil
Instructor: Pekka Heikkinen M.Sc.(Tech.)
Nokia Siemens Networks
Supervisor: Professor Raimo Kantola
Agenda
1. Background
What is Performance Based Path Determination?
2. Objectives
3. Performance Tests
Delay measurements
Throughput measurements
5. Applying PBPD
Background (1/4)

Performance of inter-process communication in distributed systems is of utmost importance


Heterogeneity in the communication needs of processes running in distributed processing
systems.



Some events require low latency communication, some other might need high bandwidth
Several different communication networks such as ATM, HIPPI, FDDI etc. available today

Each originally developed for a different application domain.

Different performance for different types of communication
Processing nodes having in-built support for multiple heterogeneous networks is becoming
common

“No
Influences the performance of the entire system
Simultaneous use of multiple networks for performance improvement not in wide use.
single network can provide the best performance for all types of communication within a single application! ”
Background (2/4)

A method of utilizing multiple heterogeneous
networks available in a distributed system to
enhance the performance of each type of
communication between processes.
Network A
Latency
What is Performance Based Path
Determination ?
Latency

Network B
Network A
Network B
t

Utilizing the difference in the
characteristics of different paths
performance
S
Message size

Two techniques :


Performance Based Path Selection (PBPS)
Performance Based Path Aggregation (PBPA)
Latency
Case X
a
Case Y
Network B
Network A
Message size
Case Z
b
Message size
Background (3/4)



Performance Based Path Selection
Applicable when one of the networks exhibits
better performances over the other(s) in one
situation, while another is better in another
situation.
fPBPS(x)
f1 (x) = a 1 x + l1
y = Latency

f2 (x) = a 2 x + l2
a1
a2
t1
t2
Dynamically select the appropriate
communication path for a given
communication event
f1 (m1 ) = t1
f2 (m2 ) = t2
t2
t1
Using some message parameters such as
message size, the type of communication
etc.
fPBPS(m1 ) = Best[f1 (m1 ),f2 (m1 )]
m1
m2
x = Message size
Background (4/4)

Performance Based Path Aggregation
f (x ) = a x + l
1



Applicable when two co-existíng networks show similar
characteristics
y = Latency

Two identical networks are aggregated into a single
virtual network
Each message divided into submessages and transferred
over available networks simultaneously.
Segmentation into submessages and reassembly at
destination can add substantial overhead.
Submessage size can be calculated by:
1
1
f (x ) = a x + l
2
a
1
a
2
2
2
f
PBPA
(x)
t =t
1
2
f (m ) = t
f (m ) = t
l2
f
PBPA
1
1
1
2
2
2
(m + m ) = t = t
1
2
l1
x = Message size
f1(x)|x= m1 = t1 = a1m1 + l1
m
1
m
2
f2(x)|x= m2 = t2 = a2m2 + l2
t1 = t 2
m = m1 + m2
Solving….
m1 
a 2 m l 2  l 1
a1 a 2
m2 
a 1 m l 1  l 2
a1 a 2
m=m +m
1
2
1
2
Objectives

Measure and compare the performance of ATM and Fast Ethernet as node inter-connects in
a distributed system.

Investigate the possibility of achieving performance enhancement in inter-process
communication in an experimental distributed system, using the techniques of Performance
Based Path Determination, when ATM and Ethernet co-exist.
Performance Tests
Tests done on a mini-network
Intel PCs connected back to back (i586 & i686)
Independent connections with Fast Ethernet and ATM
Operating System – RH Linux (2.4.18)
Ethernet NIC of type 3COM 3c905c-Tx
ATM NIC of type ENI 155p-MF
Measurement from the perspective of application programs
UDP/IP and TCP/IP protocols used over Ethernet.
AAL5 used for ATM
No load in the network or on CPU
Ethernet
NIC
ATM NIC
ATM NIC
Multimode Fiber
Cross over cable
Tested Parameters
Delay as a function of message size
Throughput as a function message
size
Ethernet
NIC
Delay measurements

Round trip delay measured
Message size range 5bytes – 60k
bytes used

Results


For a minimal size message

UDP/IP has one way delay of 63µs

TCP/IP has one way delay of 84.5

ATM has one way delay of 82.5

Below 200 bytes Ethernet with UDP has lower
delay (25% better)

For bigger messages, ATM performed better

E.g., for 2000 bytes ATM showed about
50% improvement

Delay (Contd.)

Delay breakdown

Propagation delay – negligible in our test results

Transmission delay – Significant as message size increases
 E.g., for 1000 bytes, for Ethernet this value is 80µs and for ATM 51µs
 Significant portion of total delay for 1000 bytes; 150µs and 123µs respectively

Nodal processing - Major contributor
 Found to be  29µs for sending and  65µs for receiving for Ethernet using UDP for
1000 bytes
 Most expensive - Interrupt handling and copying data to user space
Throughput measurements

Receiving throughput measured

Faster processor used as receiver

Results

Ethernet reached a maximum
throughput of 93.3Mbps with TCP

Ethernet reached a maximum
throughput 95.64Mbps with UDP

ATM reached a maximum throughput of
135Mbps

Overhead due to protocols reduces the
maximum achievable throughput

Below 1000 bytes, Ethernet offers
higher bandwidth!
Delay measurements with background traffic
800
Delay with load in the network and on the
CPU
 Multiple applications communicating
 Similar kind of traffic in the background
 Results



Performance of Ethernet degrades.
ATM has no significant impact on delay
Available throughput dropped
700
600
Delay ( microseconds )

500
400
300
TCP(40%load)
UDP(40%load)
ATM(40%load)
TCP(20%load)
UDP(20%load)
ATM(20%load)
200
100
0
500
1000
1500
Message size ( bytes )
2000
2500
Performance Tests

Conclusions from the tests

ATM has higher throughput for bigger messages and smaller delay for bigger messages
 Delay comparable to Ethernet for small messages (Depends on transport protocol)
 Better performance on a loaded network
 Overall ATM gave a better performance

We need more information, e.g., goodput ratio, connection set up time, reliability etc.
 Suitability to a system depends on the application domain, nature of inter-process
messaging of the system etc.
Applying PBPD to the test network

Performance Based Path Selection

Delay





For messages smaller than 200 bytes,
20µs improvement if Ethernet used
with UDP
And above 200 bytes, ATM behaves
better.
By dynamically selecting appropriate
path, delay performance improves
Applicability in real systems depends
on the system
Throughput



Below 1000 bytes Ethernet offered
higher throughput
For bigger messages ATM is better
Yes! Excellent possibility for
improvement!
Applying PBPD to the test network

Performance Based Path Aggregation

Implemented user process, that can
segment and sent messages over the
two paths using two threads



Delay




Used pre-determined sizes for test
purposes
Similar process for receiving
Performance degrades
Segmentation and reassembly adds
substantial overhead
Also other factors – sending and
receiving slower than for a single
network
Throughput



Significant improvement
Better than using Ethernet alone, for the
entire tested message size range
Better than ATM for above 2500 bytes
Applying PBPD

Conclusions

PBPD can offer performance improvement in some systems.

Not a surefire solution for all performance problems.

Depends on many factors including processing power of computing nodes, used
protocols etc.
Future Work

Numerous possibilities!

Cost effectiveness of PBPD


Whether additional hardware cost and R&D costs are justified by the performance
improvement
Goodput Ratio and connection setup time for ATM and Ethernet
Thank you!