DragonFly users List (threaded) for 2008-02
DragonFly BSD
DragonFly users List (threaded) for 2008-02
[Date Prev][Date Next]  [Thread Prev][Thread Next]  [Date Index][Thread Index]

Re: Dragonfly Routers

From: Bill Hacker <wbh@xxxxxxxxxxxxx>
Date: Sat, 23 Feb 2008 08:26:27 +0000

Adrian Chadd wrote:
On 20/02/2008, Bill Hacker <wbh@conducive.org> wrote:

 Routing and firewalling is a specialty that has become a very
 high-volume hardware/ASIC/RTOS field where any router a PC could at one
 time match on speed has become so cheap and flexible off-the-shelf it is
 no longer worth the bother to roll yer own *and maintain it* for any
 serious throughput.

Thing is, there are people who report doing 10ge (almost) line rate 64 byte pps on current PC hardware, with the "right" combination of PCIe, decent chipsets and crazy tuned forwarding code complete with prefetching.

Its just not being done in open source.

Nor for 'just one' 10GigE port, and only on 'PC' hardware for grins.

Serious system vendors (Cisco, Foundry, Juniper - many others) are able, for example, to set-up and tear-down multiple thousands of hardware-accelerated SSL connections *per second*, keep tens of thousands of simultaneous SSL sessions active at a time.

Or other specific routing or switching tasks - few of them on the same specialized box.

And carrier data centers have racks and racks of the various flavors of these. They have to.

The speed is due to hardware, firmware, RISC, ASIC, ... e.g. 'bespoke' silicon and single-minded software.

Not really what one would consider a proper 'OS' at all.

There may be a browser interface for configuration, but asking one of these boxes to be a general-purpose httpd, let alone an MTA or an RDBMS host, would be expensive and frustrating. Bit like swatting flies with a 16" 50. Wrong tool for the job.

And the reverse w/r 'PC' platforms.

At the 'low end', cheap-and-cheerful gear from Asian makers with 40 GB/s backplane 'fabric' now costs about the price of a decent CPU - let alone the fast NICS, RAM, and MB to support said CPU. Or the maintenance.

Once a niche has been commoditized - and cheaply - it is time to offload that function and turn to challenges NOT yet so well-covered.

Clustering and better-managed large file systems come to mind...



[Date Prev][Date Next]  [Thread Prev][Thread Next]  [Date Index][Thread Index]