View
213
Download
0
Tags:
Embed Size (px)
Citation preview
University Of Maryland 1
A Study Of Cyclone Technology
University Of Maryland 2
Table of Content
• Overview
• Contributions
• The need for time-based resource management
• Cyclone technology - basic idea
• Description of a Cyclone network
• Clock synchronization
• Data movements
• Connection management
• Scheduling
• Adaptation layer
• Fault handling
• Performance
• Advantages and limitations
• Open issues
University Of Maryland 3
• Current networking– event-based, on-demand resource allocation
– best effort performance
• New classes of traffic placing stringent requirements on the communications
• Time-based resource management in a synchronous manner
• End-to-end coordination among network components– no congestion, loss, jitter
– better utilization of bandwidth• one byte header
• reduced control messages
• reduced routing information
– well-suited network environment for traffic with stringent timing requirements
Overview
University Of Maryland 4
Contributions
• Time-based networking technology– components
– protocols
– operations
– host interface
• Time-based resource management
• Alternate way of managing resources in networking– without requiring very accurate and highly synchronized clocks
– without consuming significant amount of bandwidth for handling timing variability
University Of Maryland 5
Time-based Resource Management In Networking
l in k 1
lin k 2n o d e X
n o d eY
n o d e ZA B
C D
t1 t2
t3 t4
t5 t6
t7 t8
t9 t10
A
B
Data Loss
t1 t2
t3 t4
t5 t6
t7 t8
t9 t10
A
B
Data Loss Prevention
Delay and Jitter
A
B
t1 t2
t3 t4
t9 t10
t5 t6
t7 t8
t11 t12
Delay and Jitter Control
A
B
t1 t2
t3 t4
t9 t10
t5 t6
t7 t8
t11 t12
University Of Maryland 6
Cyclone Technology - Basic Idea
tAtA+sA
tA+2sAtA+3sA
tA+isA
tA+(i-1)sA
tA+(i+1)sA
tBtB+sBtB+2sB
tB+(j-1)sB
tB+(j+1)sB
tB+jsB
. . .
. . .
tAtA+sA
tA+dtA+sA+d
University Of Maryland 7
Data Movements - Basic
Incoming link Outgoing link
Slot buffer Pointer buffer
Free slot list
University Of Maryland 8
Data Movements
controller
switch
host
University Of Maryland 9
Types Of Traffic Supported
• Connection-oriented
• Scheduled traffic– data available at known time instant (temporal profile)
– resources reserved when establishing a connection
• On-demand traffic– source routing
– dynamic resource allocation
– possible loss of a chunk
University Of Maryland 10
Description Of Cyclone Network
• Chunk
• Slot and slot time
• Time tag
• Period
• Fixed design parameters– the size of a chunk
– the duration of a period
University Of Maryland 11
Chunk Types
• Three types - Control/Scheduled/On-demand
• Control chunk– multiple sub-chunks in a chunk
– connection request chunk
– confirm/reject/abort/terminate chunk
– pathfinder chunk
• Scheduled chunk– scheduled traffic data chunk
– scheduled traffic data acknowledgement/retransmission request chunk
• On-demand chunk– on-demand traffic data chunk
– on-demand traffic data acknowledgement/retransmission request chunk
marker checksum
University Of Maryland 12
Multiple Sub-chunks In A Chunk
marker n subchunk paddingsubchunk checksum. . .
512 bytes
8 bit marker8 bit n is the number of subchunks contained;
maximum is (512-1-1-2)/bwhere b is the number of bytes in a subchunk (127)
b*n bit subchunks where b is the number of bits in a subchunkv bit padding where (v=(512-1-1-b*n-2)*8>=0)16 bit checksum
University Of Maryland 13
Connection Request Chunk
8 bit marker160 bit source address (128 bit address + 32 bit port)160 bit destination address (128 bit address + 32 bit port)8 bit type of services64 bit start time64 bit end time16 bit s is the number of temporal descriptors in this chunk24*sbit temporal behaviorv bit padding where (v=(512-1-20-20-1-8-8-2-3*s-2)*8>=0)16 bit checksum
512 bytes
marker ToSsource destination paddingtemporalbehavior
checksumstart end s
8 bit marker160 bit source address (128 bit address + 32 bit port)160 bit destination address (128 bit address + 32 bit port)16 bit s is the number of temporal descriptors in this chunk24*sbit temporal behaviorv bit padding where (v=(512-1-20-20-2-3*s-2)*8>=0)16 bit checksum
512 bytes
marker source destination paddingtemporalbehavior
checksums
University Of Maryland 14
Confirm/Reject/Abort/Terminate Chunk
8 bit marker160 bit source address (128 bit address + 32 bit port)160 bit destination address (128 bit address + 32 bit port)672 bit padding (127-1-20-20-2)*216 bit checksum
marker source destination padding checksum
127 bytes
University Of Maryland 15
Pathfinder Chunk
8 bit marker160 bit source address (128 bit address + 32 bit port)160 bit destination address (128 bit address + 32 bit port)8 bit number of hops(h)*8 bit pointer into current hop position**8*h*2 bit list of outgoing and incoming link identifier pair at each hop*v bit padding where (v=(127-1-20-20-1-1-1*h*2-2)*8>=0)*16 bit checksum** recomputed at each hop on forward path**recomputed at each hop on forward and reverse paths
marker source destination list* checksum*hops* padding*
127 bytes
pointer**
University Of Maryland 16
Scheduled Traffic Data and Acknowledgement/Retransmission Request Chunk
8 bit chunk marker8*d bit data (d bytes of data)v bit padding where (v=(512-1-1*d-2)*8>=0)16 bit checksum
datamarker checksumpadding
512 bytes
8 bit marker160 bit source address (128 bit address + 32 bit port)160 bit destination address (128 bit address + 32 bit port)16 bit slot sequence number for this connection656 bit padding (127-1-20-20-2-2)*816 bit checksum
marker source destination sequence padding checksum
127 bytes
University Of Maryland 17
On-Demand Traffic Data And Acknowledgement/Retransmission Request Chunk
8 bit check marker8 bit number of hops8 bit index into current hop information8*h*2 bit list of outgoing and incoming link identifier pair at each hop160 bit source address160 bit destination address8*d bit data (d byte data)v bit padding where (v=(512-1-1-1-1*h*2-20-20-1*d-2)*8>=0)16 bit checksum** recomputed at each hop
marker list data padding checksum*
512 bytes
hop pointer* source destination
8 bit marker8 bit number of hops(h)8 bit index into current hop information*8*h*2 bit list of outgoing and incoming link identifier pair at each hop160 bit source address (128 bit address + 32 bit port)160 bit destination address (128 bit address + 32 bit port)8*d bit data containing acknowledgement/retransmit requestv bit padding where (v=(127-1-1-1-1*h*2-20-20-1*d-2)*8>=0)16 bit checksum** recomputed at each hop
marker source destinationlist checksum*hops padding
127 bytes
pointer* data
University Of Maryland 18
Cyclone Network ModelNon-Cyclone
Network
Non-Cyclone Network
Tem poral Regulator
Cyclonode
controller
switch
Incoming link 1
Incoming link m
outgoing link 1
outgoing link m
Backup Incoming
link 1’
Backup Incoming link m’
Backup outgoing link 1’
Backup outgoing link m’
1’
m’
host
University Of Maryland 19
Clock Synchronization
• Markers are sent to indicate the beginning and ending of a period
• A node obtains the clock information of upstream nodes
• A local clock rate is set to the average of incoming clock rates and its own rate
• A local clock phase is set considering clock phase information
Clock Adjustment At A Node With Four Incoming Links
University Of Maryland 20
Connection Establishment/Termination
A
B
C
D
E6
12
5
7
4
3
University Of Maryland 21
Time Relationship Of Scheduling
Timeline of incoming link
Timeline of outgoing link (same speed)
Timeline of outgoing link (slower)
Timeline of outgoing link (faster)
t1 t2
t3 t4
t3
t3
t4
t4
4231 tttt
University Of Maryland 22
Single-Pass Scheduling Approach
0 1 2 3 4 5 6 7
22
23
24
25
0 1 2 3 4 5 6 7
22
23
24
25
26
0 1 2 3 4 5 6 7
22
23
24
25
26
21
27
28
Available slot list
(2, 3, 5, 6, 7)
(3, 6, 7)
(3)
(1, 3, 4, 5)
• First available slot column assignment
• Assignment before reusing the buffer space
University Of Maryland 23
Double-Pass Scheduling Approach
0 1 2 3 4 5 6 7
21
22
23
24
25
26
27
0 1 2 3 4 5 6 7
28
21
22
23
24
25
26
27
Available slot list
(1, 2, 3, 4, 6)
(2, 3, 4)
0 1 2 3 4 5 6 7
28
21
22
23
24
25
26
27
Available slot list
(1, 2, 3, 4, 6)
(2, 3, 4)
University Of Maryland 24
Double-Pass Scheduling Approach
0 1 2 3 4 5 6 7
21
22
23
24
25
26
27
0 1 2 3 4 5 6 7
28
21
22
23
24
25
26
27
Available slot list
(1, 2, 3, 4, 6)
(2, 3, 4)
University Of Maryland 25
Adaptation Layer
• Receive information from an application and provide the appropriate information to a temporal regulator in the form acceptable to a Cyclone network
– Specify temporal profile for a scheduled traffic
– Provide scheduled data chunks according to the temporal profile specified
– Detect bit-errors
– Initiate appropriate recovery mechanisms
– Initiate command control chunks
– Detect data loss for on-demand traffic
• Supports existing applications and communications with non-Cyclone networks
A p p lic a t io n L a y e r
C y c lo n e A d a p t a t io n L a y e r A p p lic a t io n In t e r fa c e (C A L -A )
C y c lo n e A d a p t a t io n L a y e r C y c lo n e In t e r fa c e (C A L -C )
S w it c h
C o n t ro lle r
Hos
t
Tem
pora
l Reg
ulat
or
A p p lic a t io n L a y e r
C A L -A
C A L -C
S w it c h
C o n t ro lle r
Hos
t
Tem
pora
l Reg
ulat
or
T C P
IP
N e t w o rkIn t e r fa c e
University Of Maryland 26
Fault Handling
• Backup link approach– line condition monitor
– backup link pointer
– automatic switching to backup
• Application-dependent recovery mechanism
– primary and secondary connections
– altering partial or entirety of a path
• Reader-writer flag for handling timing variability
s lo t b u ffe r
p o in t e r b u f fe rf re e s lo t lis t
m a rk e r c h e c k e r
in c o m in glin k 1
b a c k u pin c o m in g
lin k 4
b a c k u po u t g o in g
lin k 3
o u t g o in glin k 1
controller
n e xt f re e s lo tp o in t e r
s witch
lin e c o n d it io n m o n it o r
3
in c o m in glin k 2
o u t g o in glin k 2
b a c k u pin c o m in g
lin k 3
b a c k u po u t g o in g
lin k 4
4
b a c k u p lin kp o in t e r
b a c k u p lin kp o in t e r
University Of Maryland 27
Summary Of Scheduling Techniques
Cyclone Stop-and-Go Jitter-EDD W FQBounded Delay Y Y Y YBounded Jitter Y Y Y NBandwidthProtection
Y Y Y Y
Advantages Provides bounded delay and jitterNo additional operations duringdata transm ission.No header.Easy hardware im plem entation
Provides boundeddelay and jitterBounded bufferrequirem ent
Provides boundeddelay and jitterBounded bufferrequirem ent
Sm oothes bursts withoutpolicing
Disadvantages Clock synchronizationExplicit resource reservations
Coupled delay boundand bandwidthallocation
To reduce delay jitter,all packets received alarge delayCom plicateim plementation due toseparate regulatorCostly local delaybound com putation
Coupled delay boundand bandwidthallocationExpensive roundnum ber com putation
E-to-E DelayBound PH
N
mN
HT2
D (or HH
D)
max1 SH
Jitter Bound
University Of Maryland 28
Performance Studies
• Data transfer
• Connection admission– long term connections
– single node and multiple nodes
University Of Maryland 29
Type Of Traffic Studied
• Regularly spaced traffic
• Random pattern traffic
University Of Maryland 30
Regularly Spaced Traffic (S5-S7)
0
20
40
60
80
100
1201 5 9
13
17
21
25
29
33
37
41
45
49
53
57
61
65
69
73
77
81
85
89
93
97
number of trials
nu
mb
er
of
slo
ts(S5: 5-10%) 97.33% Loading(S6: 10-20%) 94.32% Loading(S7: 5-33%) 95.24% Loading
University Of Maryland 31
0
20
40
60
80
100
1201 5 9
13
17
21
25
29
33
37
41
45
49
53
57
61
65
69
73
77
81
85
89
93
97
number of trials
nu
mb
er
of
slo
tsnumber of additionally used slots after 15 requests
number of used slots before the first rejection
(S7) 98.1% Loading
trials 26-30 of (S7)
75
80
85
90
95
100
105
1 2 3 4 5
number of trials
nu
mb
er o
f sl
ots
number of additionally usedslots after 15 requests
number of used slots before thefirst rejection
University Of Maryland 32
Random Pattern (S8)
0
20
40
60
80
100
120
0 20 40 60 80 100 120
% of available slots requested
p(a)
d-avg
University Of Maryland 33
Random Pattern (S9)
01020304050
60708090
100
0 20 40 60 80 100 120
% of available slots requested
p(a)
d-avg
University Of Maryland 34
Random Pattern (S10)
0
20
40
60
80
100
120
0 20 40 60 80 100 120
% of available slots requested
p(a)
d-avg
University Of Maryland 35
Random Pattern (S11)
0
20
40
60
80
100
120
0 20 40 60 80 100 120
% of available slots requested
p(a)
d-avg
University Of Maryland 36
Multiple Nodes
• Multiplied probability
• End-to-end delay is added
University Of Maryland 37
Summary Of Performance Issues
• A couple of millisecond per node connection establishment overhead
• Close to 100% loading for identical, regularly spaced traffic
• Above 90% loading for regularly spaced traffic
• Above 80% loading for arbitrary pattern traffic, requesting 10% of bandwidth
• More than 50% of acceptance when links are 80% loaded, requesting 10% of bandwidth
University Of Maryland 38
Summary
• Time-based resource management approach in networking
• All aspects of a computer network required to support time-based resource management
– both scheduled and on-demand traffic
– end-to-end resource usage scheduling in time
– calendar-based data movements
– existing applications and communications with non-Cyclone network
– fault condition handling
• The feasibility of Cyclone technology– end-to-end delay
– connection establishment overhead
– the probability of connection acceptance
University Of Maryland 39
Advantages
• Temporal determinacy
• Loss free and jitter free end-to-end data delivery with minimal latency, without sustaining significant delays in connection establishment
• Nearly all of the bandwidth available for the actual transmission of data
• High loading without having any adverse impact on performance
• Well-suited for hardware implementation
• Highly scalable
University Of Maryland 40
Limitations
• Temporal determinacy– temporal profile of a connection be known ahead
– not current practice in networking
• Handling applications with significant variability
• Synchronous system
University Of Maryland 41
Open Issues
• Alternative designs and policies
• Optimization of many system parameters
• Design tradeoff and optimization for specific applications
• Dynamic monitoring of performance
• Support existing internetworking protocols other than TCP/IP
• Extensions to point-to-point links and point-to-point connections
• Hardware design
University Of Maryland 42
Value Added
• Time-based resource management
• Alternate way of managing resources in networking– without requiring very accurate and highly synchronized clocks
– without consuming significant amount of bandwidth for handling timing variability
• Current applications with stringent timing requirement will perform better
• Lead to development of new classes of applications that are possible only when tight timing guarantee can be given
University Of Maryland 43
Pathfinder
D
1S
3
checksumpaddingmarker S D 0 0 4
2
2 71 1 12 152 23 3 4 15 644
4
4
2
2
7
7
12
12
15
156
6
4
4
15
15
University Of Maryland 44
Connection Establishment/Termination
A
B
C
D
E6
12
5
7
4
3