The cost comparision calculator gives you 3 different options:
Calculate the best price for cpu intense applications
Calculate the best price for memory intense applications
Calculate the best cheapest instances available.
The cost comparison calculator currently compares the following platforms:
Google Compute Engine
Rackspace Cloud Servers
Windows Azure Virtual Machines
A challenge is, what to compare. Each vendor has different instance types, so a comparison is often not so easy. Therefore, I have decided to split the comparison into different challenges as described above. For the first calculation – with cpu intense applications – i’ve used the lowest available instance. Basically, all 4 instance types matched despite Google. Google offers less memory than the other instances. The cheapest instances targets targets micro instances. All vendors despite Google offers micro instances. The memory comparison targets high memory instances. The calculation is based on available memory. The memory is aligned to the number of instances necessary. For instance, if we select 68 GB of RAM, we need one instance at Amazon EC2 but multiple instances with Windows Azure. The prices are not aligned to instances but to the sum of memory. This leads to cheaper prices when using EC2 for lower memory. I will try to adjust this in the future.
To get updates on the calculator, you can subscribe to the newsletter below.
Whenever we talk about Cloud Computing, we also talk about the postive effects it has on costs. The reason for that is simple: at first sight, Cloud Computing offers look very cheap. Most of the time, we have per-hour prices far below 1$ per hour. But the truth is that we have to keep an instance up and running for the full month in most cases. Different Cloud Computing providers offer different pricing models. This is our Guide to navigate you through the pricing models we have to deal with in the Cloud.
The easiest costs we can calculate with are fixed costs. This means that there is exactly one price for a month. The price consists of the monthly fee multiplied by the service unit. A service unit is a valuation which is most of the time the number of users. If you consume a service in your company and you need 10 users for this service, the service unit is the user. Software as a Service (SaaS) is the most common usage for this. Let’s assume you rent an e-mail service in your company. Your company consists of 10 people, where each person needs an e-mail service. The price per user is 5$ per month. Summing it up, you need to pay 50$ per month for your e-mail service. The formular, in short:
Service Charges per month x service units.
Whereas Software as a Service Solutions are very easy, Platform as a Service (PaaS) and Infrastructure as a Service (IaaS) are getting more complicated. What
we can see here is a more flexible cost approach, where you have to pay on different topics. These costs are common with the big players in Cloud Computing such as Amazon, Google, Microsoft and Rackspace. We have to deal with several different structures. At the core of each platform you might rent an Instance. Each Instance runs for a certain number of hours, named Compute Hours. Those Platforms often require storage, e.g. for Media Data (Images, Videos) or a Databases. To calculate storage, the providers charge on a per-GB price. This price has to be multiplied by the number of GB stored. Since PaaS and IaaS Solutions consume Network and Internet Traffic, we also get charged on Incoming and Outgoing Bandwidth. Here applies the same as with storage: the charge-ability is based on a per-GB fee. Last but not least, Storage Transactions also fall into this topic. A Storage Transaction is an operation you do on Objects or Datasets such as deleting, creating, updating or reading data. Most of the time, this is measured via the ReST-Keywords. The formular in short:
Costs per Compute Hour x Compute Hours+ Costs per GB Stored x GB Stored
+ Costs per Incoming Bandwidth x Incoming Bandwidth
+ Costs per Outgoing Bandwidth x Outgoing Bandwidth
+ Sum of Storage Transaction x Costs per Storage Transactions
Hybrid Costs are a mixed form of the former two cost structures, where you find a variable part and a fixed part. This doesn’t occur too often at all but it makes sense in some cases, since it adds possibilities to save money. Currently, this model is offered by Amazon Web Services (AWS) with the name “Reserved Instances”. The key is that you pay a certain amount upfront and your hourly fee gets reduced. Normally, the Break-Even-Point for this Cloud Computing pricing model is reached after 6 month or more. So if you target your instance to run for more than 6 month, you might look into that.
Often talked about and offering some great benefits, but what are flexible costs? Normally flexible costs are referred to as “spot instances”. This means that you enter a certain amount for an instance that you are willing to pay for. If the spot price is lower than the amount you entered, your instance is ready to use. This pricing model is currently offered by Amazon Web Services (AWS). Spot prices gives us more flexibility on our costs, since your instances are only available if the price is at a certain level. Especially startups or universities will find this very interesting.
Having a closer look at pricing models in the cloud gives greater flexibility and enables us to save money. There are also more advanced technologies that also help to save money such as newvem, what we will talk about later.