Hi Ben, Sorry for the slow response - took a while to catch up on emails after SC14. Yes, I would still need a 4x10GbE public connected resource in JLSE. For the spec, can we go with the recommendation provided here? https://fasterdata.es.net/science-dmz/DTN/hardware-selection/ Thanks, Raj On Nov 24, 2014, at 3:43 PM, Allen, Benjamin S. <[email protected]> wrote:

Hi Raj,

Operations has been discussing what todo with the Tubes testbed so it more appropriately sits in one of JLSE's public network subnets.

We're interested the timeline of returning the DTN hardware that makes up Tubes back to production, so I want to explore what hardware we have in JLSE that could be reasonably updated to provide a similar resource if needed.

First do we still need a 4x10GbE public connected resource in JLSE?

If we're going to re-use JLSE hardware or build a replacement, what would your ideal "Tubes" testbed be spec'ed like, CPU, memory, etc. Note, we don't have 40GbE switch ports, so I can only replace the machine with another that has 4x10GbE either connected through JLSE's switches or directly to Donald (ALCF's core router in 240).

Thanks,

Ben