OpenSplice DDS Finance

OpenSplice DDS
Meeting Skyrocketing
Message Rates in Complex
Financial Trading Applications
Angelo Corsaro, Ph.D.

angelo.corsaro@prismtech.com
Product Marketing Manager | PrismTech
OMG RTESS and DDS SIG Co-Chair
Agenda
OpenSplice DDS
‣ Current Challenges
‣ OpenSplice DDS
‣ Technology Ecosystem
‣ DDS TouchStone
‣ Use Cases Dr. Angelo Corsaro
‣ Concluding Remarks
© 2008, PrismTech. All Rights Reserved

Historical Perspective
Once upon a time

‣ Trading was an activity entirely centered
around humans
‣ Trading rates were limited by human speed
From 1980’s to 1990’s
‣ 1985 signed the birth of digital trading and
the surge of data aggregators
‣ Humans increasingly assisted by machines
From early 2000’s to today
‣ The rise of completely automated
electronic trading and exchange
‣ Direct connection to stock exchange
‣ Decimalization and exceptional market
growth leading to rise in data volumes

Trends & Challenges - Financial Firms
Trends
‣ Direct connection to exchanges is increasingly
chosen over data aggregators to reduce trading
latency
‣ Market data is growing steadily
‣ Milliseconds speedup over competitors can be
worth millions of dollars
Challenges
‣ Process several hundreds-of-thousands of
messages, quickly approaching millions
messages, per second
‣ Minimize latency and jitter, while maintaining the
system stability and reliability
Deployed middleware solutions were not
designed to address these challenges!

Trends Challenges - Stock Exchanges
Trends
‣ More and more firms connecting directly
‣ Increasing number of sources providing the
same quotes
‣ Competitive advantage obtained by
reducing latency on data feeds
Challenges
‣ Ensuring fairness of data distribution
‣ Minimizing latency
‣ Scaling with the number of destinations
Deployed middleware solutions were not

designed to address these challenges!

Resulting Technical Challenges
Power is nothing without control!
We need a Topic-Based Publish/Subscribe Middleware that provides...
Most Hyped
Low latency and predictable data dissemination
Support for several 100K messages/sec, up to
Millions messages/sec

Resulting Technical Challenges
Power is nothing without control!
We need a Topic-Based Publish/Subscribe Middleware that provides...
Most Hyped
Support for several 100K messages/sec, up to
Millions messages/sec
Not so Hyped, yet Fundamentally Important
Stability under overload condition
Scalability
Fairness
Control over latency/throughput tradeoffs
Traffic Engineering
HW Filtering
Traffic Shaping
Priority driven delivery
High Performance Persistence
Event Processing
Agenda
OpenSplice DDS
‣ OpenSplice DDS
‣ DDS TouchStone
‣ Use Cases

OpenSplice DDS
The future of Publish/Subscribe has come!
A High Performance Real-Time Data-Centric Publish/Subscribe Middleware

‣ The right data, at the right place, at the right time -- all the time!
‣ Fully distributed, multicast-enabled, high performance, highly scalable, and high availability,
hot-swap, architecture
Perfect Blend of Data-Centric and Real-Time Publish/Subscribe Technologies

‣ Content based subscriptions, (continuous) queries, filters, and windows
‣ Fine grained tuning of resource usage and data delivery and availability QoS
‣ Optimal networking and computing resources usage
Loosely coupled Subscriber
Publisher
‣ Plug and Play Architecture with Dynamic
Discovery B
‣ Time and Space Decoupling

m
A F Subscriber
Open Standard Publisher J
D C
‣ Complies with the full profile of the OMG K Subscriber

DDS v1.2
E
Publisher
Subscriber
Global Data Space

OpenSplice DDS: Foundational Abstractions
‣ Information Model. Defines the structure, relations, and QoS, of the

information exchanged by the applications, and supports Flat, Relational, and
Object Oriented Modeling
‣ Typed Global Data Space. A logical data space in which applications read and
write data anonymously and asynchronously, decoupled in space and time
‣ Publisher/Subscriber. Produce/Consume information into/from the Global
Data Space
‣ QoS. Regulates the non-functional properties of information in the Global Data
Space, e.g., reliability, availability, and timeliness, etc.
Subscriber
Publisher
A F Subscriber
Publisher J
D C
K
E
Subscriber
Publisher
Global Data Space Subscriber

Flat Modeling
How do I define the data and subscribe to it?
‣ Modeling. Items distributed within the

system are modeled as Topics
‣ Topics. Defined as the associations between

a data type and a set of QoS and identified
by a unique name.
‣ Data Types. The data type associated to a

Topic is specified by means of IDL
‣ Subscriptions. Topics, Content-Filtered

Topics or Multi-Topics (join between topics) struct StockQuote {
can be used to specify subscriptions. SQL string symbol;
string name;
expression are used to specify filters and sting exchange;
joins float quointe;
string xml_extra;
};
StockQuote
symbol: "GOOG"
name: "Google Inc."
StockQuote
StockQuote exchange: "NASD"
symbol: "MSFT"
symbol: "AAPL" quote: 663.97
name: "Microsoft Corp."
name: "Apple Inc." xml_extra: "<ext>,,,</ext>
exchange: "NASD"
exchange: "NASD" quote: 33.73
quote: 165.37 xml_extra: "<ext>,,,</ext>
xml_extra: "<ext>,,,</ext>

Relational Modeling
How do I define the data and subscribe to it?
‣ Modeling. As in a Relational DB, an

OpenSplice DDS information model can be
represented by means of Entity Relationship
(ER) diagrams
‣ Topics. The entities, represented by means of

Topics, are in turns an association between a
data type and a set of QoS and identified by a
unique name (like tables in an RDBMS) struct StockQuote {
string symbol;
‣ Data Types. The data type associated to a string name;

sting exchange;
Topic must be a structured type expressed in float quote;
IDL };
string xml_extra;
#pragma keylist StockQuote symbol

‣ Instances. Key values in a datatype uniquely StockQuote
identify an instance (like rows in table) symbol: "AAPL"
name: "Apple Inc."
StockQuote
symbol: "MSFT"
exchange: "NASD" name: "Microsoft Corp."
quote: 165.37 exchange: "NASD"
‣ Correlation. SQL Expressions can be used to

xml_extra: "<ext>,,,</ext> quote: 33.73
correlate information by means of key values StockQuote
symbol: "GOOG"
name: "Google Inc."
exchange: "NASD"
quote: 663.97
Object Oriented Modeling
Unleashing the power of Objects...
The DDS supports true Object Oriented Distributed Person

Information Modeling by means of the DLRL layer ssn: string
age: short
‣ Reduced Complexity & Improved Productivity name: string
surname: string
‣ Focus on the architecture and business logic, while hiding away birth_date: string
sex: char
the details involved with the diffusion of shared objects state getX()
‣ Encapsulation setX()
‣ Attributes are only accessible through dedicated getter/setter

operations, i.e., don’t need to the messaging middleware or the Employee
title: string
application to have privileged access to business objects job_code: string *
representation salary: long
‣ Local Operations
getEmployer()
setEmployer()
‣ Besides getters/setters, all other kind of manipulations can be is employed employs

done using custom operations Employer
‣ Inheritance 1
Attribute
Attribute
Operation
‣ Single inheritance supported for DLRL Objects Operation
‣ Navigable Relationships
‣ Single Relationships
‣ Multi Relationships (Set, Map, List)
Relational/OO Mapping
OO ➡ Relational
Canvas
handle: Handle Object Oriented
‣ Middleware can automatically Flight
RadarTrack
manage the generation and FlightView
image: string
call_sign: string
route: list
x: float
association between the refresh_rate: int * 1 origin: string 1
y: float
1 z: float
Object-Oriented Model and the visible: bool
uid: GUID
dest: string
radar_id: string
aircraft: string
Relational Model display() operationX()
operationOne()
operationTwo()
operationY()
FlightViewTopic FlightViewAssociation
FlightTopic Relational
call_sign: string
image: string call_sign: string
route: list RadarTrackTopic
refresh_rate: int uid: GUID
origin: string x: float
visible: bool handle: Handle
dest: string y: float
uid: GUID
aircraft: string z: float
radar_id: string radar_id: string
Relational ➡ OO
‣ The Relational Model can be mapped to an Object Oriented model
‣ The mapping is under control of the architect

Multiple Mappings
Object Oriented
FastFlightView
image: string
Canvas
handle: Handle Object Oriented
refresh_rate: int
visible: bool Flight
RadarTrack
uid: GUID FlightView call_sign: string
x: float
x: float image: string route: list
y: float
y: float refresh_rate: int * 1 origin: string 1 1 z: float
z: float visible: bool dest: string
radar_id: string
call_sign: string uid: GUID aircraft: string
operationOne()
display() display() operationX()
operationTwo()
operationY()
FlightViewTopic FlightViewAssociation
FlightTopic Relational
call_sign: string
image: string call_sign: string
route: list RadarTrackTopic
refresh_rate: int uid: GUID
origin: string x: float
visible: bool handle: Handle
dest: string y: float
uid: GUID
aircraft: string z: float
radar_id: string radar_id: string
Multiple Mappings
‣ Allow different, Object-Oriented, local reconstruction of the same underlying relation model
‣ The state is shared across all the local reconstruction

Processing Content & Structure
Complex Event Processing...
‣ Content-based Subscriptions OpenSplice DDS some of the most

useful features found in Complex
‣ Topics Joins & Projections Event Processing platforms!
‣ Queries
‣ Continuous Queries
‣ Correlation across topics (via joins)
‣ Events windows (via History QoS)
StockQuote
symbol: "AAPL"
S1
name: "AppleStockQuote
Inc."
symbol:
exchange: "AAPL"
"NASD"
name:
quote: 578.19 "Apple Inc."
StockQuote
exchange:
symbol:"NASD"
name:
quote:
"MSFT"
"Microsoft
387.74 Corp."
StockQuote
{symbol = “AAPL”}
exchange:symbol:
"NASD""GOOG"
StockQuote
quote: 33.73
name: "Google
symbol:Inc."
"AAPL"
exchange: "NASD"
name: "Apple Inc."
quote: 663.97
exchange: "NASD"
quote: 165.37
S2
P {quote > 500}
S3
{ name =”Microsoft Corp.”}

Processing Content & Structure
Complex Event Processing...
‣ Content-based Subscriptions OpenSplice DDS some of the most

useful features found in Complex
‣ Topics Joins & Projections Event Processing platforms!
‣ Queries
‣ Continuous Queries
‣ Correlation across topics (via joins)
‣ Events windows (via History QoS) StockQuote
symbol: "AAPL"
name: "Apple Inc."
exchange: "NASD"StockQuote
symbol: "AAPL"
quote: 578.19
StockQuote
name: "Apple Inc."
S1
symbol: "AAPL"
exchange: "NASD"
quote: 387.74name: "Apple Inc."
exchange: "NASD"
quote: 165.37 {symbol = “AAPL”}
StockQuote
symbol: "AAPL"
name: "Apple Inc." StockQuote
symbol: "GOOG"
exchange: "NASD"
quote: 578.19 name: "Google Inc."
S2
P
exchange: "NASD"
quote: 663.97
{quote > 500}
StockQuote
symbol: "MSFT"
exchange: "NASD"
quote: 33.73
S3
{ name =”Microsoft Corp.”}


information exchanged by the applications, and supports Simple, Relational,
and Object Oriented Modeling
Data Space
Subscriber
Publisher
A F Subscriber
Publisher J
D C
K
E
Subscriber
Publisher

Global Data Space
How can I organize my data?
‣ The Global Data Space can organized into domains which in turns
can have partitions
‣ Each partition is mapped to an IP Multicast Address
‣ QoS controls the availability and consistency of data
IP Multicast per Partition

Hardware Filtering
B Traffic Confinement
A F
J
D C
K
Partition E
Information Segregation
Traffic Confinement


Data Space
Subscriber
Publisher
A F Subscriber
Publisher J
D C
K
E
Subscriber
Publisher

Publisher/DataWriter & Subscriber/DataReader
Publisher/DataWriter Subscriber/DataReader
‣ Publishers are responsible for managing the ‣ Subscribers are responsible for managing the
dissemination of publications reception of data resulting from subscriptions
‣ The dissemination is driven by the QoS ‣ The presentation of data is driven by the QoS
associated with the DataWriter, the Publisher, associated with the DataReader, the Publisher
and the Topic and the Topic
‣ A DataWriter is associated with only one ‣ A DataReader is associated with only one
Publisher and one Topic. It embeds the Publisher and one Topic. It embeds the
knowledge of dealing with the Topic’s Data knowledge of dealing with the Topic’s Data Type
Type
Type Matching
QoS matching
QoS QoS QoS QoS QoS
Topic
Name
Publisher Subscriber
... DataWriter writes Type reads DataReader

...
...
DataWriter writes Type reads DataReader
Name
Topic
QoS QoS QoS
Network

Data Space
Subscriber
Publisher
A F Subscriber
Publisher J
D C
K
E
Subscriber
Publisher

QoS Model
How is QoS matched in the System?
‣ QoS can associated with all relevant OpenSplice DDS entities

‣ Some QoS are matched based on a Request vs. Offered Model
‣ Publications and Subscriptions match only if the declared and
requested QoS are compatible
‣ e.g., it is not possible to match a publisher which delivers data unreliably with a subscriber
which requires reliability
Type Matching
QoS matching
QoS QoS QoS QoS QoS
Topic
Name
Publisher Subscriber
... DataWriter writes Type reads DataReader

...
...
Name
Topic
QoS QoS QoS

QoS
What are the key QoS supported?
QoS Policy Applicability RxO Modifiable QoS matching
DURABILITY T, DR, DW Y N Data QoS QoS QoS
DURABILITY T, DW N N Availability Topic QoS

QoS
SERVICE
Name
LIFESPAN T, DW - Y Publisher ...

Subscriber
HISTORY T, DR, DW N N DataWriter writes Type reads DataReader
Name
PRESENTATION P, S Y N Data Delivery Topic
RELIABILITY T, DR, DW Y N QoS QoS QoS
PARTITION P, S N Y QoS matching
DESTINATION T, DR, DW Y N
ORDER
OWNERSHIP
OWNERSHIP
T, DR, DW
DW
Y
-
N
Y
‣ Rich set of QoS allow to
STRENGTH configure several
T, DR, DW Y Y
DEADLINE
LATENCY T, DR, DW Y Y
Data
Timeliness
different aspects of data
BUDGET availability, delivery and
TRANSPORT T, DW - Y
PRIORITY timeliness
‣ QoS can be used to
TIME BASED DR - Y Resources
FILTER
RESOURCE
LIMITS
T, DR, DW N N control and optimize
USER_DATA DP, DR, DW N Y Configuration network as well as
TOPIC_DATA T N Y
computing resource
GROUP_DATA P, S N Y

Mapping QoS
Which properties does QoS controls?
TimeBasedFilter Deadline
Data
Throughput
Latency
LatencyBudget TransportPriority
Control over Latency/Throughput tradeoff

Control over data latency
Control over data priority

Mapping QoS
TimeBasedFilter Deadline History
Data Data
Throughput Lifespan Durability
Latency Availability
Ownership
LatencyBudget TransportPriority Ownership
Strength

Control over data queueing
Control over data persistency
Control over data sources hot-swap

Mapping QoS
TimeBasedFilter Deadline History
Data Data
Throughput Lifespan Durability
Latency Availability
Ownership
LatencyBudget TransportPriority Ownership
Strength

Control over data queueing
Control over data persistency
Control over data sources hot-swap
Reliability
OpenSplice DDS provides
Destination
programmatic QoS-driven
Presentation Data Delivery
Order support for configuring the
Control over data distribution reliability
most important properties of
Control over data ordering data distribution!
Control over presentation

TimeBasedFilter Deadline
Data
Throughput
Latency
LatencyBudget TransportPriority
Data Timeliness
QoS Policy Applicability RxO Modifiable

DEADLINE T, DR, DW Y Y
LATENCY T, DR, DW Y Y
Data
BUDGET
Timeliness
TRANSPORT T, DW - Y
PRIORITY

Deadline
You can’t be later than...
The DEADLINE QoS policy allows to define the maximum inter-

arrival time between data samples
‣ DataWriter indicates DEADLINE T, DR, DW Y Y
that the application

commits to write a new
QoS matching
value at least once

QoS QoS QoS
every deadline period QoS Topic

Name
QoS
‣ DataReaders are
Publisher ... Subscriber

... ...
notified by the DDS DataWriter writes Type reads DataReader
when the DEADLINE Name

Topic
QoS contract is
violated QoS QoS QoS
QoS matching

Latency Budget
I need to get there in at most...
The LATENCY_BUDGET QoS policy specifies the maximum

acceptable delay from the time the data is written until the data is
inserted in the receiver's application-cache
‣ The default value of the QoS Policy Applicability RxO Modifiable

duration is zero LATENCY T, DR, DW Y Y
BUDGET
indicating that the delay
should be minimized QoS matching
‣ This policy is a hint to the

QoS QoS QoS
DDS, not something that QoS Topic

Name
QoS
must be monitored or Publisher

Subscriber
... ...
enforced. ...
Name
Topic
QoS QoS QoS
QoS matching

Transport Priority
VIP Data, stay clear!
The TRANSPORT_PRIORITY QoS policy is a hint to the infrastructure

as to how to set the priority of the underlying transport used to send
the data.
TRANSPORT T, DW - Y
PRIORITY
QoS matching
QoS QoS QoS
QoS Topic QoS

Name

... ...
Name
Topic
QoS QoS QoS
QoS matching

OpenSplice DDS Architectural Outlook
Scalable & Flexible
Application Application Application Application

OpenSplice DDS OpenSplice DDS OpenSplice DDS OpenSplice DDS
Binding Binding Binding Binding
Shared Memory
OpenSplice DDS OpenSplice DDS OpenSplice DDS OpenSplice DDS OpenSplice DDS
Binding Binding Binding Binding Binding
Config. SOAP Networking Durability DBMS

Security

Networking Architecture
Getting control over the network!
Architecture Shared Single Copy per Node Shared

Pack Across Topics/Applications
‣ Network-channels Memory Optimal Unmarshaling Memory
‣ Priority bands
‣ Network-partitions
‣ Multicast Groups
‣ Traffic-shaping OpenSplice DDS OpenSplice DDS
‣ Burst/Throughput Binding Binding
Pre-emptive Network Scheduer
Priority Scheduler
Networking Data Urgency Traffic Pacing
Networking
Scalability and Efficiency
‣ Single shared library for Network Channels
Priority Bands
applications & services
‣ Ring-fenced shared memory Traffic Shaping
segment
‣ Data urgency driven network-
packing
Determinism & Safety
‣ Preemptive network-scheduler
‣ Data importance based network-channel selection
‣ Partition based multicast-group selection
‣ Managed critical network-resource

Head of Line Blocking
If a single connection is used for all the data, High Priority Data can experience
unbounded priority inversion -- also known as head-of-line blocking
This is a know problem of non priority preserving transport such as TCP-UDP/IP
75 55 15 10
Priority Bands ensure that high priority data, always find its way preempting
lower priority data
With priority bands, priority inversion is completely under control of the

Engineer that design/deploys the system
75 90 90
55 60 60
15 10 30 30

History
Data
Lifespan Durability
Availability
Ownership
Ownership
Strength
Data Availability
DURABILITY T, DR, DW Y N
DURABILITY T, DW N N
SERVICE Data
Availability
LIFESPAN T, DW - Y
HISTORY T, DR, DW N N

Durability
For who will this data be available?
The DURABILITY QoS controls the data availability w.r.t. late joiners,
specifically the DDS provides the following variants:
‣ Volatile. No need to keep QoS Policy Applicability RxO Modifiable
data instances for late DURABILITY T, DR, DW Y N

DURABILITY T, DW N N
joining data readers SERVICE
‣ Transient Local. Data

instance availability for late
QoS matching
joining data reader is tied to

QoS QoS QoS
the data writer availability QoS Topic

Name
QoS
‣ Transient. Data instance DataWriter writes Type reads DataReader
availability outlives the data Publisher ... ... ... Subscriber
writer DataWriter writes Type reads DataReader
‣
Name
Persistent. Data instance Topic
availability outlives system QoS QoS QoS
restarts QoS matching
The DURABILITY_SERVICE QoS provide control over configuration of the

service that implements the transient and persistent durability features
Persistence Architecture
Architecture Shared Memory Shared Memory

‣ Fault-Tolerant Data
Availability
‣ Transient -- on memory
‣ Persistent -- on disk OpenSplice DDS OpenSplice DDS
Disk
OpenSplice DDS OpenSplice DDS
Binding Binding Binding
‣ Partitioning
Binding
‣ DDS Partitions Networking Durability Durability Networking

Persist Partitions
‣ Alignment Persistent Data on Disk
Transient Data in Memory
‣ Dedicated Channels
Dedicated Persistence Service Alignment Channel
Goal Features
‣ Transient QoS. Keep state-data ‣ Fault-tolerant availability of non-volatile data
outside the scope/lifecycle of its
publishers
‣ Efficient delivery of initial data to late-joining
applications
‣ Persistence QoS. Keep ‣ Pluggable Durability Service
persistent settings to outlive the
system downtime ‣ Automatic alignment of replicated durability-services
Lifespan
For how long will this data be available?
‣ The LIFESPAN QoS

LIFESPAN T, DW - Y
policy allows to
control what happens
QoS matching
to stale data QoS QoS QoS
‣ It specifies the validity Topic

interval for data
QoS QoS
Name
written by the

... ...
DataWriter
‣ The default validity Topic

Name
interval is infinite
QoS QoS QoS
QoS matching

History
How many data samples should I keep?
The HISTORY QoS policy controls whether the DDS should deliver only the most
recent value, attempt to deliver all intermediate values, or do something in
between. The policy can be configured to provide the following semantics:
‣ Keep Last. The DDS will only

attempt to keep the most QoS Policy Applicability RxO Modifiable
HISTORY T, DR, DW N N
recent “depth” samples of
each instance of data QoS matching
identified by its key QoS QoS QoS
‣ Keep All. The DDS will

QoS Topic QoS
attempt to keep all the Name
samples of each instance of DataWriter writes Type reads DataReader
data identified by its key.

... ...
‣ On the DataWriter Samples DataWriter writes Type reads DataReader
are kept until delivered to all Topic

Name
known subscribers
QoS QoS QoS
‣ On the DataReader side QoS matching

samples are kept until the
application “takes” them

Reliability
Destination
Presentation Data Delivery
Order
Data Delivery
PRESENTATION P, S Y N
RELIABILITY T, DR, DW Y N
PARTITION P, S N Y
DESTINATION T, DR, DW Y N Data Delivery
ORDER
OWNERSHIP T, DR, DW Y N
OWNERSHIP DW - Y
STRENGTH

Reliability
How much effort should be taken to deliver data?
The RELIABILITY QoS indicate the level of guarantee offered by

the DDS in delivering data to subscribers. Possible variants are:
‣ Reliable. In steady-state
the middleware QoS Policy Applicability RxO Modifiable
RELIABILITY T, DR, DW Y N
guarantees that all
samples in the QoS matching
DataWriter history will QoS QoS QoS
eventually be delivered QoS Topic

Name
QoS
to all the DataReader DataWriter writes Type reads DataReader
‣ Best Effort. Indicates

... ...
that it is acceptable to DataWriter writes Type reads
Name
DataReader
not retry propagation of Topic
any samples QoS QoS QoS
QoS matching

Ownership
Who owns the data?
The OWNERSHIP QoS specifies whether it is allowed for multiple

DataWriters to write the same instance of the data and if so, how
these modifications should be arbitrated. Possible choices are:
‣ Shared. Multiple writers are

allowed to update the same QoS Policy Applicability RxO Modifiable
instance and all the updates OWNERSHIP T, DR, DW Y N
are made available to the QoS matching

reader QoS QoS QoS
‣ Exclusive. Indicates that Topic

each instance can only be
QoS QoS
Name
owned by one DataWriter, but Publisher

Subscriber
... ...
the owner of an instance can ...
change dynamically -- due to DataWriter writes Type reads DataReader
liveliness changes
Name
Topic
‣ The selection of the owner is QoS QoS QoS
controlled by the setting of QoS matching
the
OWNERSHIP_STRENGTH
Ownership Strength
How strong are you?
The OWNERSHIP_STRENGTH Specifies the value of the “strength”

used to arbitrate among DataWriters that attempt to modify the same
data instance
‣ Data instance are QoS Policy Applicability RxO Modifiable
identified by the OWNERSHIP

STRENGTH
DW - Y
couple (Topic, Key)

‣ The policy applies QoS matching
only if the
QoS QoS QoS
OWNERSHIP is QoS Topic

Name
QoS
EXCLUSIVE DataWriter writes Type reads DataReader

... ...
Name
Topic
QoS QoS QoS
QoS matching

Hot-Swap & Hot-Hot

Managing Hot-Swap
‣ A Topic can have Shared or Exclusive Ownership

‣ Exclusively owned Topics can be modified by a single writer
‣ Writer strength is used to coordinate replicated writers
StockQuote
symbol: "MSFT"
StockQuote
exchange:symbol:
"NASD" "GOOG"
quote: 33.73
name: "Google Inc."
exchange:StockQuote
"NASD"
P
quote:symbol:
663.97"AAPL"
name: "Apple Inc."
exchange: "NASD"
S1
quote: 165.37
STRENGTH=3
StockQuote
symbol: "AAPL"
StockQuote
name: "Apple Inc."
symbol:exchange:
"GOOG""NASD"
name: quote:
"Google Inc."
165.37
StockQuote
exchange: "NASD"
symbol: "MSFT"
quote: 663.97
exchange: "NASD"
quote: 33.73
P’ S2
{quote > 500}
STRENGTH=2 StockQuote
symbol: "AAPL"
StockQuote
name: "Apple Inc."
symbol:exchange:
"GOOG""NASD"
name: quote:
"Google Inc."
165.37
StockQuote
exchange: "NASD"
symbol: "MSFT"
quote: 663.97
exchange: "NASD"
quote: 33.73
P’’ S3
{ name =”Microsoft Coorp.”}
STRENGTH=1

Managing Hot-Swap

StockQuote
symbol: "MSFT"
StockQuote
exchange:symbol:
"NASD" "GOOG"
quote: 33.73
name: "Google Inc."
exchange: "NASD"
P
quote: 663.97
StockQuote
symbol: "AAPL"
S1
STRENGTH=3 name: "Apple Inc."
exchange: "NASD"
quote: 165.37
StockQuote
symbol: "GOOG"
name: "Google Inc."
StockQuote
exchange: "NASD"
symbol: "MSFT"
quote: 663.97
exchange: "NASD"
quote: 33.73
P’ S2
{quote > 500}
STRENGTH=2 StockQuote
symbol: "GOOG"
name: "Google Inc."
StockQuote
exchange: "NASD"
symbol: "MSFT"
quote: 663.97
exchange: "NASD"
quote: 33.73
P’’ S3
STRENGTH=1

Managing Hot-Swap

StockQuote
symbol: "MSFT"
exchange: "NASD"
quote: 33.73
P
StockQuote
symbol: "AAPL"
S1
STRENGTH=3 name: "Apple Inc."
exchange: "NASD"
quote: 165.37
StockQuote
symbol: "MSFT"
exchange: "NASD"
quote: 33.73
StockQuote
P’ S2
symbol: "GOOG"
name: "Google Inc."
exchange: "NASD"
quote: 663.97
{quote > 500}

STRENGTH=2
StockQuote
symbol: "MSFT"
exchange: "NASD"
quote: 33.73
P’’ S3
STRENGTH=1

Managing Hot-Swap

‣ Similar behavior, could be achieved by relying on write_with_timestamp
S1
StockQuote
symbol: "AAPL"
name: "Apple Inc."
exchange: "NASD"
quote: 165.37
StockQuote
S2
symbol: "GOOG"
name: "Google Inc."
exchange: "NASD"
quote: 663.97
{quote > 500}

StockQuote
symbol: "MSFT"
exchange: "NASD"
quote: 33.73
P’’ S3
STRENGTH=1

Managing Hot-Swap

‣ Similar behavior, could be achieved by relying on write_with_timestamp
S1
StockQuote
symbol: "AAPL"
name: "Apple Inc."
exchange: "NASD"
quote: 165.37
StockQuote
S2
symbol: "GOOG"
name: "Google Inc."
exchange: "NASD"
quote: 663.97
{quote > 500}
StockQuote
symbol: "MSFT"
S3
P’’
exchange: "NASD"
quote: 33.73

STRENGTH=1

Managing Hot-Hot Data Replication
‣ Hot-Hot replicas could be used to ensure that only, and always, the
fastest sample is delivered
‣ This can be achieved in a few different manners, the most effecive
being a combination of DESTINATION_ORDER QoS and
write_with_timestamp()
StockQuote
symbol: "MSFT"
StockQuote
exchange:symbol:
"NASD" "GOOG"
quote: 33.73
name: "Google Inc."
exchange:StockQuote
"NASD"
P
quote:symbol:
663.97"AAPL"
name: "Apple Inc."
exchange: "NASD"
S1
quote: 165.37
T1 Data Feed {symbol = “AAPL”}
StockQuote
symbol: "AAPL"
StockQuote
name: "Apple Inc."
symbol:exchange:
"GOOG""NASD"
name: quote:
"Google Inc."
165.37
StockQuote
exchange: "NASD"
symbol: "MSFT"
quote: 663.97
exchange: "NASD"
quote: 33.73
P’ S2
{quote > 500}
Satellite Data Feed StockQuote
symbol: "AAPL"
StockQuote
name: "Apple Inc."
symbol:exchange:
"GOOG""NASD"
name: quote:
"Google Inc."
165.37
StockQuote
exchange: "NASD"
symbol: "MSFT"
quote: 663.97
exchange: "NASD"
quote: 33.73
P’’ S3
Another Data Feed
Agenda
OpenSplice DDS
‣ OpenSplice DDS
‣ DDS TouchStone
‣ Use Cases

OpenSplice™ DDS
Monitor & Control Logging & Replay Configuration High-Level

Services
C/C++ Java XML
DLRL
OMG DDS
DCPS
Ownership Persistence Content Subscription

Profiles
Minimum Profile
DDS API
SOAP DBMS Durability Security Networking Pluggable
Services
OS RTOS Supported
[Linux, Windows, Solaris) [VxWorks, RT-Linux, LynxOS) Platforms

OpenSplice DDS PowerTools
Simply Productivity...
Domain Specific Language

User
Interface
Remote Information Application Deployment

Tuner Plug-ins
Connect Modeling Modeling Modeling
Eclipse Foundation

OpenSplice DDS PowerTools
End-to-End System Design cycle

‣ Information, Application, and Deployment
Modeling
‣ Productive and correctness-enforcing modeling
environment
Information Modeling
‣ Graphical system-wide Information and QoS
Modeling
‣ OpenSplice DDS code-generation
‣ Support for libraries of reusable Information
Model
Application Modeling
‣ Graphical application modeling
‣ Pattern-oriented code-generation (listener/
waitset/MVC)
Deployment Modeling
‣ Modeling of DDS-configuration Service
configuration (networking, durability)
‣ Runtime control (& round-trip engineering) by
OpenSplice DDS Tuner™

MessageBoard Example – Productivity
605
228
150
67
62 % Relational 71 %
OO & 123 %
reduction Modeling reduction increase SQL
MDE
Relational & MDE
Modeling
ODBC/SQL
MDE: a factor of 10 !

Integration Technologies
51All Rights Reserved

© 2008, PrismTech.
DBMS-Connect™ Technology
Connecting to the Enterprise World...
Features
Application Application
‣ Transparently ‘connect’
OpenSplice DDS ‘Real-time
OpenSplice DDS OpenSplice DDS Application information-space’ to the enterprise
Binding Binding DBMS space
JDBC/ODBC 3.0
Binding ‣ Share/access data in the most
‘natural way’ between Real-Time
and Enterprise applications
‣ Standards based (ODBC/JDBC)
Shared Memory and optimized performance for the
DBMS
ANTs™ data-server
‣ Seamless two-way data
exchange
‣ OpenSplice DDS persistence-
profile support for QoS-driven
OpenSplice DDS OpenSplice DDS OpenSplice DDS JDBC/ODBC 3.0
Binding Binding Binding Binding replication/persistence, even for
DBMS users
Config. Networking DBMS
‣ Highly configurable for
maximum performance

Web Service Gateway
Expose you data to the WWW

Service
Management
SOAP SOAP (Dynamic)

Shared Memory
Interface DDS-API’s
service
OpenSplice DDS OpenSplice DDS OpenSplice DDS OpenSplice DDS OpenSplice DDS
Configuration & Binding Binding Binding Binding Binding
Status Reporting Config. SOAP Networking Durability DBMS
Pluggable Service Framework Web Services Interface

‣ Service-management. Automatic start/stop, ‣ XML-based SOAP interface (remote
liveliness-monitoring, restart policy ‘connect’ via HTTP/URL)
‣ Maintenance. Common status and error ‣ XML-based Discovery of DDS-entities:
reporting Topics/Partitions, Subscribers/Readers,
‣ Configuration. XML-based configuration with Publishers/Writers
matching deployment tooling (‘OpenSplice ‣ XML-based Type-support: runtime creation
DDS configurator’) of topics/readers/writers
‣ Dynamic OpenSplice API’s. Untyped ‣ XML-based C&M: control and monitor of
readers/writers, dynamic type-support, C&M any DDS-entity (QoS, Resource-usage),
(discovery/control/monitoring) especially for OpenSplice Tuner™

Secure Networking
Securing you communication Infrastructure
 Channel encryption between nodes

secure information distribution
over untrusted networks
between nodes of different security level Shared Memory
enforcement of need-to-know principle
 Integrity and data origin authentication for

information distribution over the network OpenSplice DDS OpenSplice DDS OpenSplice DDS OpenSplice DDS OpenSplice DDS
Binding Binding Binding Binding Binding
Config. SOAP Networking Durability DBMS

Security
A dedicated crypto channel per network Infrastructure solution

partition ‣ Security enforcement part of the
‣ Ensures that information received via the information backbone (transparent to
network (or eavesdropped on the network) applications)
can only be retrieved in its RED ‣ Secured Infrastructure
(unencrypted) form on nodes that are
accredited or have a need-to-know Secure networking is first building block for
complete IA
‣ Authentication of data origin
‣ Integrity preservation of information ‣ End-to-end security, incl. mandatory access
control per application, over the networked
exchanged between nodes
DDS backbone is next

Agenda
OpenSplice DDS
‣ OpenSplice DDS
‣ DDS TouchStone
‣ Use Cases

Perspectives on Performance
Let’s make it easy, let’s make it real!
Everyone is Giving Numbers...

‣ Performance data abounds in Messaging Solutions Datasheets
‣ How meaningful is to provide that data?!?
‣ Depending on configuration and HW platform, OpenSplice DDS can
ship from several hundreds of thousands, to millions, of messages
per second, with latencies in the order of 100us...
‣ So What? What does this really buy you?
The Real Stuff
‣ Instead of being told about numbers...
‣ How about a benchmarking framework that allows with a few
clicks to set-up a benchmark representative of your application
that you can run on your lab, or even target system?
‣ Isn’t that much better than just numbers?

DDS TouchStone
Measure what you care, where you care!
‣ OpenSplice DDS ships with DDS TouchStone, a Benchmarking Framework that

allows to set-up, with a few clicks, a benchmarking application and to measure
latencies and throughputs
‣ All the most important characteristic of an application can be controlled, such as
data size, generation frequency, QoS, scheduling, and middleware
configuration
‣ DDS TouchStone provides a time-effective and meaningful way of assessing
OpenSplice DDS suitability for a specific appliation

Agenda
OpenSplice DDS
‣ OpenSplice DDS
‣ DDS TouchStone
‣ Use Cases

OpenSplice DDS Beyond Finance
Defense
‣ TACTICOS-CMS: THALES Naval Netherlands’
CMS, 26 ships classes, >100 ships
‣ > 2.000 deployed runtimes (running on Solaris-servers, Linux-
consoles, and vxWorks embedded subsystems)
‣ 15 Navies worldwide (Netherlands, Germany, Turkey, Greece,
Oman, Qatar, Portugal, South Korea, Japan, Poland,…)
‣ USA programs: LCS/GD, ENFMC/NG, LHA-

LHD/DRS
‣ Brazilian Navy
‣ Australia: DSTO, ADI (Australia)
‣ THALES Naval NL’s Flycatcher system
‣ 4 army’s, >400 deployments
‣ NSWC: Open Architecture Test Facility (OA-TF)
Tactical networks
‣ Ultra Electronics (US, UK): OpenSplice DDS
selected over competition for superior
scalability and fault-tolerance

OpenSplice DDS Beyond Finance
Transportation
‣ Amsterdam Metro
‣ Paris Metro
‣ CoFlight: Flight-plan management system upgrades
for France, Italy, Switzerland
Aerospace
‣ NASA Kennedy Space Center: Constellation
Program for next generation ARES Rocket Launch
System
SCADA
‣ Chemtech/Siemens in Brazil: since 2006
Homeland Security
‣ IDA: ‘Cybercrime Defense’ in WAN environment

Agenda
OpenSplice DDS
‣ OpenSplice DDS
‣ DDS TouchStone
‣ Use Cases

Why OpenSplice DDS?
Giving you Power and Control!
Most Hyped
Support for several 100K messages/sec, up to Millions messages/sec
Not so Hyped, yet Fundamentally Important
Stability under overload condition
Scalability
Fairness Optimally addresses the requirements of next
Control over latency/throughput tradeoffs generation Financial Application
Traffic Engineering
HW Filtering
Traffic Shaping
Priority driven delivery
High Performance Persistence
Event Processing
Other
Security
DBMS/Web Services Connectivity
Concluding Remarks
Applicability
‣ OpenSplice DDS uniquely addresses some of the key requirements for next
generation Financial Applications
‣ Thanks to its architecture, OpenSplice DDS delivers extremely high
performance, high availability and reliability, while retaining low and
predictable latency even under burst of activities
Open Architecture
‣ OpenSplice DDS is the only implementation in the world which fully
implements the OMG DDS v1.2 standard
Technology Ecosystem
‣ Seamless cohabitation with CORBA (Java+C++) and RTSJ
‣ DBMS Integration
‣ Security Plug-in
OpenSplice DDS is the best solution available on the market for

solving next-generation Financial Application challenges!

Contact Us
‣ OpenSpliceDDS Resource Center

‣ http://www.prismtech.com/opensplice-dds/
‣ Evaluate OpenSplice DDS

‣ Training and Consulting
‣ sales@prismtech.com
‣ OMG DDS Information

‣ http://www.dds-forum.org/
‣ http://portals.omg.org/dds/
Thank You! © 2008, PrismTech. All Rights Reserved

OpenSplice DDS Finance

Uploaded by

Copyright:

Available Formats

You might also like

OpenSplice DDS Finance

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

OpenSplice DDS Finance

Uploaded by

Copyright:

Available Formats

OpenSplice DDS

Angelo Corsaro, Ph.D.

© 2008, PrismTech. All Rights Reserved

Once upon a time

© 2008, PrismTech. All Rights Reserved

© 2008, PrismTech. All Rights Reserved

Deployed middleware solutions were not

© 2008, PrismTech. All Rights Reserved

We need a Topic-Based Publish/Subscribe Middleware that provides...

© 2008, PrismTech. All Rights Reserved

We need a Topic-Based Publish/Subscribe Middleware that provides...

© 2008, PrismTech. All Rights Reserved

A High Performance Real-Time Data-Centric Publish/Subscribe Middleware

Perfect Blend of Data-Centric and Real-Time Publish/Subscribe Technologies

‣ Time and Space Decoupling

‣ Complies with the full profile of the OMG K Subscriber

Global Data Space

© 2008, PrismTech. All Rights Reserved

‣ Information Model. Defines the structure, relations, and QoS, of the

© 2008, PrismTech. All Rights Reserved

‣ Modeling. Items distributed within the

‣ Topics. Defined as the associations between

‣ Data Types. The data type associated to a

‣ Subscriptions. Topics, Content-Filtered

© 2008, PrismTech. All Rights Reserved

‣ Modeling. As in a Relational DB, an

‣ Topics. The entities, represented by means of

‣ Data Types. The data type associated to a string name;

#pragma keylist StockQuote symbol

‣ Correlation. SQL Expressions can be used to

The DDS supports true Object Oriented Distributed Person

‣ Attributes are only accessible through dedicated getter/setter

‣ Besides getters/setters, all other kind of manipulations can be is employed employs

© 2008, PrismTech. All Rights Reserved

© 2008, PrismTech. All Rights Reserved

‣ Content-based Subscriptions OpenSplice DDS some of the most

© 2008, PrismTech. All Rights Reserved

‣ Content-based Subscriptions OpenSplice DDS some of the most

© 2008, PrismTech. All Rights Reserved

‣ Information Model. Defines the structure, relations, and QoS, of the

© 2008, PrismTech. All Rights Reserved

IP Multicast per Partition

© 2008, PrismTech. All Rights Reserved

‣ Information Model. Defines the structure, relations, and QoS, of the

© 2008, PrismTech. All Rights Reserved

QoS QoS QoS QoS QoS

... DataWriter writes Type reads DataReader

QoS QoS QoS

‣ Information Model. Defines the structure, relations, and QoS, of the

© 2008, PrismTech. All Rights Reserved

‣ QoS can associated with all relevant OpenSplice DDS entities

QoS QoS QoS QoS QoS

... DataWriter writes Type reads DataReader

QoS QoS QoS

© 2008, PrismTech. All Rights Reserved

DURABILITY T, DR, DW Y N Data QoS QoS QoS

DURABILITY T, DW N N Availability Topic QoS

DataWriter writes Type reads DataReader