[Atlas-anchors-pilot] Provisional hardware specs for Atlas Anchors

[Daniel Karrenberg]

I like the active discussion!

Let me say a few things to explain the motivation:

Our experience with TTM has been that the lack of a replacement cycle is very painful because maintaining and operating really outdated kit is just a very painful thing to do. I asked the team to spec a machine to run multiple services quite possible isolated in VMs, but I could not say exactly which services yet. I asked the team to spec a machine that is available and supported widely in our service region. I asked the team to spec a machine that can be operated remotely with confidence and at a reasonable cost.

So  we now got what I asked for. But none of this is set in stone. We are listening.

A few comments:

Dell/HP: To our knowledge Dell is more widely available in our service region and we have a working relationship with them. It may be a sub-optimal choice locally, but globally it is the optimal one as far as we know. We will look into others if there is a widespread preference among pilot participants. We do not want a hardware zoo because operational cost dominates.

Too big: if this is a widespread concern, an alternative could be to have a small box that will support only one service at a time. For the pilot phase I would like to avoid introducing that in order to keep things manageable. Afterwards it is something well worth to consider. However even the Atlas Anchor needs disk for logging traces of the received the measurement traffic, some capacity to run target services and additional capacity to run a probe. The last thing we want is to have saved 1000.- on the hardware and having to invest a multiple of that to make the services fit.

Detailed tech specs / power economy: We are listening and the engineers will consider everything you have to say.

Daniel