What were we thinking? The C13 Outlet. Really?

Something has to be done. We all know the C13 outlet provides the vast majority of the power connectivity for IT gear across data centers worldwide.  We see these outlets and matching plugs on nearly every server, switch and storage element produced over the past dozen years. Sure there are high power plugs beginning to be seen in dense processing gear (like the L6-30), but in the bigger picture, those are not the lion’s share of power in the racks across anyone’s data center… The C13 is ubiquitous and at the same time known as a poor answer, lacks intelligence, and stifles most of the innovation possible going forward, and worst of all: creates operational nightmares in real data center operations.

The C13 Outlet

The C13 Outlet needs a replacement

What am I talking about? I have just seen a new ‘sleeve’ which is intended to make the C13/C14 physically tighter. Kinda cool answer to a silly problem. Most people know the C13/C14 connection is a loose-fitting arrangement that typically sees the spring-tension of the metal wipers within carrying the bulk of the mating integrity. Really? Data Centers are much more complex than that. Vibrations and thermal stress makes those spring-things almost useless so every manufacturer has dreamed up their own variation of a mechanical securing technology BAND-AIDs to get around the real problem. The new ‘sleeve’ is just one approach. There is a variety of pushbutton-Button versions of the outlet, some of which grab the ground pin, but costs are high and concern over long-term reliability may exist,  so usage is low.  Some vendors have a price-book full of wire tension bars that attempt to meet the soft plastic C14 plug and secure it. There are rails and bars that allow nylon tie-wraps to secure the plugs. I’ve even seen duct tape used in extreme situations. Obviously we are all wrestling with a known problem that affects the majority of the data centers worldwide. And in today’s reality we all “play with the hand we are dealt”, BUT we need a plan for the future! We shouldn’t be content on carrying forward a mechanical issue just because it has always been done this way. Again, I suggest we need a plan.

We also have a desperate need for an intelligent scheme to allow devices to communicate physically upstream. In the world of DCIM, wouldn’t it be great if the outlet itself had a method of communicating structured operational data (like energy consumption or sleep state, serial and model numbers or temp, etc). If each outlet had the ability to communicate data, the promise of DCIM would be realized. Audit would be electronic and auto-discovery would be a snap.

Conceptually, visualize a twist locking receptacle, (which implies something round), with a keyed orientation, and perhaps a 1-wire data path pin in the center. Something like a L6-30, (but perhaps a bit smaller to mimic the foot-print of a C13) and with a center communications pin. Make it public domain, and define a protocol that communicates over the one-wire path. Probably extend the PMbus work already in progress, but include system info as a primary goal. Perhaps build on a CLP-type command/response protocol so that the devices themselves could report (when asked) about their intelligence capabilities.

C13/C14 is a problem. We all know it. We all need a replacement and it has to start somewhere. It’s a long haul, a long journey, but it doesn’t start until WE START IT! The world of DCIM would be enabled by it and everybody smiles. Anyone else in?

About admin

Mark Harris Fremont, CA
This entry was posted in Obstacles and tagged , . Bookmark the permalink.

2 Responses to What were we thinking? The C13 Outlet. Really?

  1. rfeldhaus@tripplite.com says:

    Interesting article and approach, but why not report your power data over the existing network management platform and wiring. Network managers already view network management interfaces to monitor all reported aspects of their network hardware — the limitation here with power data, is that few servers at present detect and report this information. Given the greening of IT as time goes on and a demand for detailed power consumption information, your network gear will eventually report these power related items directly, but it’s unlikely that the industry will come-together to invent an entirely new pathway that transmits this information over your power lines, they’re going to use existing data pathways.

    In the meantime, you can track power consumption at the individual outlet level today with network PDU systems that include individual outlet level metering. PDU brands don’t communicate back to the NMS via the data-enriched power connections you envision, they communicate via SNMP, just like everything else. You can set tolerances for logging or notification as whole-PDU, breakered load bank or individual outlet power level threshholds are met.
    rfeldhaus@tripplite.com

    • markharris2000 says:

      You are absolutely right that power metrics should be reported over the standard production network. The main point of this post however was to identify a MECHANICAL issue with the C13 connector itself. This connector is used for the vast majority of data center devices, and yet people use everything from duct tape to sticky glue to keep it physically installed. I have seen tie-wraps, wire-brackets, and even hot glue used to keep it together. The mechanical spec for the C13 is simply a poor spec.
      (The idea of adding a data wire to some form of new locking connector was really just icing on the cake…. if a data wire DID exist, a great deal of additional information could be passed to the production network attached monitoring and control systems. Power supply health, temperatures, status, current draw, power factor, etc.)

Leave a Reply