 |
|
|
|
 |
|
| Products |
|
|
|
|
|
|
|
|
| DART |
|
|
|
|
|
|
|
|
|
|
|
Internet
Delivery
Platform
iDP |
|
|
|
 |
|
|
|
|
|
RtWebCache server pumps real-time data to client applications on
the Web. The rtWebCache server is a multi-user real-time cache and
data pump capable of supplying any type of data to clients the Web.
Metering on a per-exchange basis is integrated into the server.
Some of the benefits of using RtWebCache are:
|
 |
Accelerates requests for
market data from the Web |
|
|
Multi-user gateway may
stream updates to multiple clients on the Web or to the trading
floor |
|
|
A cost-effective alternative
for high-priced real-time middleware which is tuned to the
efficient delivery of market data information |
The server is normally configured to tie into a client's existing
infrastructure and proprietary systems - Triarch or TIB real-time
distribution platforms - and provides data from all existing systems
to the Web through a single channel with a well-defined protocol.
Two types of instruments reside in the cache:
|
1. |
Non-preempt-able instruments
specified by the server administrator. These commonly used
instruments are always available from the server and are never
pruned from the cache |
|
2. |
Preempt-able instruments
which are delayed on demand. These are removed using a least-recently
used pruning algorithm |
The server has both snapshot and update channels. The snapshot channel
is designed for Web-based queries such as from a CGI application.
Requests for multiple items on the snapshot channel are batched
together as one multi-item request and returned as one multi-item
response. The server terminates the channel after the response is
written to the client. The update channel is designed for trader
workstation applications requiring streaming, real-time data.
The multi-threaded design of the server allows it to scale with
client usage. Each real-time subscription channel such as Triarch
or TIB ciServer runs in its own thread. A client thread handles
each snapshot and update client request. This frees the real-time
threads to service the potentially high updating real-time data
stream.
A configurable pool of client threads is created at server startup.
If the number of clients exceeds the pool, then the same client
thread may service two or more clients. The server ensures that
client requests are evenly spread across all available client threads.
Snapshot responses are written to the client by the client thread;
Updates are copied from the real-time thread to an inter-thread
event queue, and then driven to the client by the client thread.
|
|
