The present and future of post production business and technology

Apologies for news feed outage today

The news feeds I create are hosted hosted by the Digital Production BuZZ. This website is hosted on Media Temple’s Gridserver, which we chose because it allows us to absorb large amounts of traffic by simply using more resources on the Gridserver (for which we pay of course).

Unfortunately, Media Temple had a major problem with the Gridserver this morning and despite multiple phone calls to tech support and their media relations people did not get returned. Finally just now, I reached Marketing Director Alex Copehart, who could comment on the record. He had a team of people with him for the call (I suspect they were already together working on understanding this morning’s problems) and I was able to ask why there wasn’t more information on the website and why I hadn’t had my calls returned.

Here’s a tip, if you get to an outside call center for an organization like this, try hitting the directory number for sales! Most companies are more responsive in sales than anywhere else. It was via sales that I was able to get a useful comment.

The problem with the lack of detailed and timely information, Alex explained, was because t they were still attempting to understand fully what went wrong and they did not want to post misleading information.

David also offered, unprompted, to find us a different solution that would be more robust as a discounted price and we’ll explore that further with him over the next few days.

In any technology “stuff happens”. The Grid Server is a new concept and one that has potential, and we’re treating this as a one-off glitch. We may get hooked up with a different solution with them or
we’ll find a more reliable hosting service. We hope that today’s problems are transient. We’ll see.

My take-away, is for businesses that are having a problem, is to be as open and honest as you can be as quickly as you can be. My biggest frustration today is that I had no idea whether this was a problem that would be fixed in an hour, in which case we’d just wait it out, or whether it was a really big problem and I’d better make other arrangements to keep our site and feed alive. We’ll certainly be exploring failover provisions for the future.






4 responses to “Apologies for news feed outage today”

  1. Greg

    Hey Phillip,

    Every time I upload to youtube my video looks like
    crap, so I was very excited to read and follow your
    advice. I did everything you recommended and…. still
    looks like crap. My piece was shot on a panasonic
    dvx100a at 24p mode, cut on fcp hd 4.5 and is 1:55

    Here it is on youtube

    Here it is on .mac

    I’d be glad to buy your download if you can help me.

  2. This may be a case of adjusting your expectations to reality :). That is Sorenson Sparke/Flash 7 that YouTube use and they blow it up after compressing it down to 320 x 240. So, if you’re already sending them the high quality “master” as per my notes then that’s probably the best you can expect.

    The parts that are really breaking up are the wireframe animations and that’s to be expected – those frames are going to be very, very hard to encode. Usually you could use 2 pass VBR and allocate some data from the easier-to-encode parts of the video (which did come out well on YouTube) to these more difficult frames. Unfortunately YouTube does not use 2 Pass VBR and that’s where the problem is coming for your material.

    YouTube is not going to give a better result than this, unless they improved the codec to Flash 8/9 (On2 VP6) and did 2 Pass VBR encoding. There are no announced plans for any change to YouTube’s encoding so this is pretty much as good as it will be for the immediate future.

    Your best solution would be to replace the wireframe animation with something a little more “encoding” friendly.


  3. Philip – did the problem with the gridservers get sorted? I’ve been with mt for years just switched to the gridservers two weeks ago…no problems so far.

  4. Hi Peter

    The original problem got resolved. We’ve been chasing down another issue where some people experience an incomplete download of the BuZZ show files.

    We thought it had been cured but we had another report this week. So we’re going to move all the media files for the BuZZ audio over to Dreamhost where we also have an account.

    MT have been helplful in trying to track down the cause of the incomplete download but really need to be able to track a “case” to an IP address and time. Then they could see what happened in their logs.

    Mostly pretty helpful and the Grid has really settled down in the last 5-6 weeks.