Cloudoscope - a cost profiler for the cloud
I am very happy to announce that I am joining CodeValue. CodeValue, as our site says, is the home for software experts. Indeed, the company is built of a group of consultants and we provide mentoring and consulting in the areas of architecture, software development and technology. While I am going to be doing some consulting, the main reason for me to join (and my main role - as VP product delivery) is the other business area of the company - building cutting edge products for developers. Which brings me to the topic of this post Cloudoscope(tm) the first in out line of products.
Cloud computing brings the era of consumption based pricing and promisses to transform computing to a utility like electricity, water etc. The pay-as-you go model brings elasticity and cost savings since, for example, you don't need to stock computing power for peak loads. When the need arise you can just add more instances etc. etc.
Cloud computing also means that your cost structure change for example on Windows Azure a service bus connection would cost you between 2$ and 4$ (500 connection @995$, 1@ 3.99$ as mentioned here) - it is easy to see how many connections you use, but how many connections do you really need?; Another example an Amazon RDS (which gives you MySQL capabilities) extra large (high memory) reserve instance will cost you $1325 (1-year) + $0.262 per hour + $0.1 Gb/month + $0.1 per million requests. On the other hand an extra large high memory EC2 instance (on which you can install mySQL) will cost you $1325 (1-year) + $01.17 per hour (plus you need to figure how to persist the data) - what will be cheaper based on your usage scenario?; partial compute hours are billed as full hours, booting a new instance takes a few minutes, if your instance is idle and based on the way your system is used what's cheaper shutting the instance down (delete the deployment in azure) or leaving it running?
Cloudoscope is our effort to try to get these types of questions answered and put you in control of your total cost of ownership. Here are some of the features we're looking at (from Alon Fliess CodeValue's CTO blog):
- Provide correlation between code and cost
- Cut total cost of ownership and save money
- Show the cost of each function and relevant line of code
- Show the cost of business requests
- Show cost improvement or degradation after a code change
- Provide optimization advices
- Provide guidance to Cost Oriented Development™
- Help trading service quality Vs. cost
- Provide a framework for developing Cost Oriented Unit Tests™
- Cost oriented cloud computing standard approval