Mogulus: Yep…It’s Still a Beta Product

This is a cautionary tale about a jaw droppingly fabulous technology that simultaneously was quite a frustrating and shaky experience as a producer with what turned out to be just over 94% uptime today. This story also illustrates how inventive and innovative technologies are really pushing the envelope with cloud computing.

Over the last several weeks, I invested days in due diligence on video streaming eventually signing up with Mogulus, “…the most powerful live broadcast platform on the internet.” While I don’t doubt they’ll achieve that vision someday, the tiny gray label in their logo that says “Beta” means that they’re actually, “…the most powerful live broadcast platform on the internet in beta.“

Any non-online software company will flat out tell you NOT to use beta software “in production”. Of course, Google’s permanent affixing of the word “beta” under Gmail means that beta has become a convenient excuse when something goes wrong and, frankly, makes you and I as users partly to blame (which reminds me of that line in the movie Animal House when, after destroying the car belonging to the brother of Flounder, one of the fraternity pledges, Otter tells him, “Hey! You fucked up…you trusted us!”

I spent all day today running an online channel filled with several dozen videos, organized into storyboards, for a 24 hour internet broadcast channel for a client. Luckily, the 28 minutes during the live, streaming conference event for my client worked flawlessly, but at least five other times during the day the streaming channel simply went black (though the ticker at the bottom of the screen worked but no video played) and four other times there was stuttering or videos repeating parts of themselves.

Frantic calls from other team members ensued, I emailed support and it corrected itself in minutes (varying from 5-8 minutes) so the email replies were always, “Ahh…I looked at it and it looks fine.”  Each time this happened (or the several times the video stuttered or replayed short snippets of a video 2 or 3 times), we lost viewers which was the real irritant.

Mogulus uses Amazon EC2 and S3, their computing and storage components in Amazon Web Services. While I’d notice in my browser status bar the words, “Read mogulus-user-files.s3.amazonaws.com” and then “Transferring from mogulus-user-files.s3.amazonaws.com” when the next video in the queue would cue up to play, that’s the only evidence I could see of a stall without any analytics.

Who was at fault? Mogulus or Amazon? Some bandwidth provider? Kittens on the loose? Doesn’t matter. What matters is the effectiveness of the service.

The bottom line? For the Pro service at $350 per month plus $1.50 per gigabyte over the 25GB’s they give you each month (and trust me…you can spend ALOT on bandwidth if you have even a modestly successful event of 100 people) I expect a lot more.

On the Mogulus forum was this analysis of just how much one could spend that is spot-on and may give you pause about doing anything mission-critical yet with Mogulus:

• So you’re paying $1.50 per gigabyte for bandwidth on Mogulus Pro. They include 25GB in their base plan at $350/mo.

• If we’re streaming at the basic 500 Kbps, that amounts to 3.75 MB/min or 225 MB/hour. So if we had 11 people watch us for 10 hours, that’s 225MB x 11 x 10 = 25 GB gone.

• If we have 1000 people watching for one hour, say a Christmas parade, that will amount to 225 GB. At $1.50 per GB, that will cost us $337.50 for that hour.

• ONE person watching for 24 hours would eat up 5.4GB ($8.10)

• If we were to average about 1000 viewers over a 24 hour period (with it peaking mid-day and petering off at night) that would amount to 5400
  GB, or $8,100. FOR ONE DAY. It would be a quarter million a month to operate at that level.

For THAT kind of money this service must be as close to perfect as possible. Again, it was really good, but in the last eight hours (9am-4pm) there were at least nine “material” outages or “stuttering” events of at least several minutes each.

If we were to consider an average of 3 minutes per outage times 9 outages, that’s just over a 94% uptime. While that’s OK for a free service, when my startup client is coughing up $350 a month plus bandwidth costs, it needs to be over 99%.

All that said, the Mogulus service is incredible. What they’ve built is unbelievably powerful and I will continue to monitor their progress closely. But at the end of the day it’s the outcome that matters FOR EVERYTHING from software to cars to your job….everything. The outcome for us was the viewer experience and it was a bit too far from rock solid for my comfort (you should probably go out and buy Maalox futures since I downed a fair amount today!).

All in all it was a successful event. There’s a loop of 3 hours of videos running now on a feature they call “Auto Pilot” and the stream has only stopped dead once for 4 minutes (I’m writing this post as I watch to make certain it keeps playing). I think I can stop for the day now and keep my fingers crossed another frantic mobile phone call doesn’t come.