Hardware accelerated messaging

June 21, 2008

Marc asked for hardware accelerated messaging ideas. Here’s one: how about using SCRAMNet to eliminate IP networking and achieve sub microsecond latency ?  You’d need all of your exchange connectivity, pricing and order management processes on boxes in the same machine room linked by a SCRAM fibre loop. But it would go mighty quick. SCRAM pricing is surprisingly reasonable too…


One Response to “Hardware accelerated messaging”

  1. Ken Says:

    Cool I’d never heard of SCRAM.
    FYI we’re beating the numbers indicated on this website my a substantial margin with Wombat Data Fabric, which is built on this same reflective memory concept. We’re talking north of a million messages per second (thats at a healthy 200byte or so payload size) across an IB or Ethernet switch, and multiple millions of seconds when sharing memory locally within a node (at sub 5 micros). And all this is over commodity parts in software (using standard I/O hardware (NICs and HCAs) for acceleration) with a generic messaging API (MAMA.) Oh and it’s in production 🙂 (the remote sharing is. Local sharing is going Beta)

    Holler if you want to know more or kick the tires… kbarnes@nyx.com

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s