Cisco Cisco Nexus 5010 Switch White Paper
of 12
W H I T E P A P E R
I n T E R - P A R T y L A T E n c y M A n A g E M E n T
of 12
W H I T E P A P E R
I n T E R - P A R T y L A T E n c y M A n A g E M E n T
different platforms to be compared. They
are rarely viewed as impartial, and they do
not reflect the full end-to-end latency that
traders experience.
are rarely viewed as impartial, and they do
not reflect the full end-to-end latency that
traders experience.
2. Inter-Party Latency
Management
Confronting these challenges leads us to
propose an inter-party latency management
(IPLM) solution that market centers, providers
and participants can jointly use to monitor
latency and performance across their respective
systems. By adopting a joint approach to
latency management, all parties can overcome
the technical obstacles that currently hinder
end-to-end visibility. Additionally, this approach
facilitates transparency between parties and
promotes collaborative problem-solving.
propose an inter-party latency management
(IPLM) solution that market centers, providers
and participants can jointly use to monitor
latency and performance across their respective
systems. By adopting a joint approach to
latency management, all parties can overcome
the technical obstacles that currently hinder
end-to-end visibility. Additionally, this approach
facilitates transparency between parties and
promotes collaborative problem-solving.
These benefits translate into strategic
advantages for each segment of the trading
loop. Traders achieve superior execution quality
and increased trading success, through better
insight into trading latencies and their impact
on algorithm performance. Market centers
attract greater liquidity from latency-sensitive
clients by increasing end-user confidence.
Market providers enhance the value of their
services by adding trader-to-venue performance
visibility, and remove any risk of being seen as
slow. All parties gain by reducing the time and
effort spent pin-pointing and fixing inter-party
latency performance problems.
advantages for each segment of the trading
loop. Traders achieve superior execution quality
and increased trading success, through better
insight into trading latencies and their impact
on algorithm performance. Market centers
attract greater liquidity from latency-sensitive
clients by increasing end-user confidence.
Market providers enhance the value of their
services by adding trader-to-venue performance
visibility, and remove any risk of being seen as
slow. All parties gain by reducing the time and
effort spent pin-pointing and fixing inter-party
latency performance problems.
3. Secure, Scaleable Latency
Measurement
Measuring latency for messages travelling
between two parties involves generating
information about when and where each
message is seen, and then exchanging this
information so that transit times can be
determined. The solution we propose is entirely
passive – we do not modify or delay messages in
any way, and the exchange of measured
information occurs out-of-band. Our solution
operates as a stand-alone system attached to
each party’s network, and requires no integration
with existing systems or protocols.
between two parties involves generating
information about when and where each
message is seen, and then exchanging this
information so that transit times can be
determined. The solution we propose is entirely
passive – we do not modify or delay messages in
any way, and the exchange of measured
information occurs out-of-band. Our solution
operates as a stand-alone system attached to
each party’s network, and requires no integration
with existing systems or protocols.
Clearly the exchange of measurement data
between parties must be secure. We use a
policy-based approach to security, giving each
party full control over who is allowed to access
data, how they are required to authenticate, and
which data they have access to. A party can
securely peer with multiple opposite parties
while ensuring that each of them has access
only to data relating to their own messages.
between parties must be secure. We use a
policy-based approach to security, giving each
party full control over who is allowed to access
data, how they are required to authenticate, and
which data they have access to. A party can
securely peer with multiple opposite parties
while ensuring that each of them has access
only to data relating to their own messages.
Exchanging latency data will obviously consume
network bandwidth. Since trading parties are
often connected via expensive WAN links, we’ve
designed our data protocols to be very
lightweight. The added overhead is less than 4%
of the traffic being measured. This ensures that
latency measurement can run alongside trading
and market data over the network infrastructure
already in place, with no requirement for new
bandwidth or special treatment.
network bandwidth. Since trading parties are
often connected via expensive WAN links, we’ve
designed our data protocols to be very
lightweight. The added overhead is less than 4%
of the traffic being measured. This ensures that
latency measurement can run alongside trading
and market data over the network infrastructure
already in place, with no requirement for new
bandwidth or special treatment.
We’ve also designed our latency measurements
to be precise. Trading parties who are located in
the same city can achieve sub-millisecond
latencies while those who are co-located can
reach the microsecond level. Our solution will
measure latency in each direction across any
to be precise. Trading parties who are located in
the same city can achieve sub-millisecond
latencies while those who are co-located can
reach the microsecond level. Our solution will
measure latency in each direction across any
Market providers enhance
the value of their services by
adding trader-to-venue
performance visibility