Mosso Pricing Update – Compute Cycles
A couple of months ago, Mosso the cloud based hosting provider introduced a change to their billing model that had existing users up in arms. The change that caused all the concern was a move to bill users partly on the number of requests that hosted sites incurred. If your sites clocked over 3 million requests in a month, you would be billed at quite a hefty rate for the number of requests over this limit. This site alone would burn 3 million requests easily given the amount of traffic I get: It only amounts to about 70000 unique visits per month.
Well the latest news is that Mosso appear to have been listening to users and revised the strategy yet again, to a metric that will be based on the concept of ‘compute cycles’. Pricing and the amount of cycles you get in your package has not been released yet but obviously users are keen to see if Mosso still provides the value for money that was originally apparent.
The fact is that with new technology models, new pricing models must follow: it remains to be seen if new pricing models build successful sustainable businesses.

May 15th, 2008 at 4:20 pm
I am also a Mosso customer and I have to say I couldn’t be more happy that they are changing their metric. This is a good move on Moss’s part. They listen to its users and reacted accordingly.
This new model goes along the same lines that Media Temple already have going. So it’s nothing new. It’s seems to be a proven metric that works and people can work with.
Cheers.
June 23rd, 2008 at 1:05 pm
I’m new to Mosso. We put up a Magento (online store) based site and we’ve maxed out our compute cycles for the sites standard package by manually adding 300 product. We haven’t even launched the store yet.
July 26th, 2008 at 12:08 am
Compute cycles is a terrible metric. I have one local website and one national website and I burned 8,000 compute cycles in 8 days. The allotment is 10,000 cycles/per standard account for 100 bucks. They estimate that I’ll be 283% over that. What is that $383 a month. I can get a dedicated box on Rackspace for that.
August 12th, 2008 at 4:10 am
Can anyone explain what is compute cycle in mosso.com, how does one compute cycle is evaluated
August 13th, 2008 at 5:20 pm
According to some Wiki I just looked at, a compute cycle consists of 2 parts: the instruction prefetch to load the instruction for execution and the execution of the instruction. However, Mosso is going beyond that by also examining disk IO so that database intensive work adds more to the compute cycles.
My partner and I have just setup a wesite for a client on Mosso’s hosting cloud and I am monitoring the cycles pretty closely since the website is rather database (MySql) intensive. Here is an example of the account usage:
Disk Usage: 12.8 MB
Bandwidth: 70.0 MB
Compute Cycles: 108
Requests: 11941
Some of that usage was on uncompiled .net code and had rapid cycle increases during a short period of time. We improved it by compiling the code but the cycles on compiled code are probably going to exceed what we are willing to accept once we find a reasonable alternative.
January 10th, 2009 at 12:54 am
Hello,
I had a bad experience with Mosso compute cycle pricing.
Everything else is stellar, the bandwidth, the ease, the price seems great … however mosso decided to take a big fat sh*t on the compute cycles.
This was for a site I made using hand written PHP/MySQL with very highly optimized code.
I wrote my own process watcher, etc. the compute cycles to me seem ‘fake’ because I did a quick test on a new site one test had heavy code, I ran that script 500 times, checked the compute cycles, next day I ran simple hello.htm page that contained “Hello” 500 times. the compute cycles were both the same, for every 50 hits = 1 compute cycle.
Very strange stuff, they don’t reveal anything about it other than generalities, your images etc … I set 100% of the images to s3.
I set php alternative caching to all the pages, set it very high to 5 days, still compute cycles ran through the roof.
So a warning, charging on something that is not verifiable is sketchy.