BitSwarm cloud as the Infrastrucutre-as-a-service for a Software-as-a-service provider

A potential customer contacted us with the question about the possibility of partnership as they would like to offer their own Software-as-a-service (SaaS) on top of our Infrastructure as a Service (Iaas) cloud. After we learned that the customer already understands the concept of the cloud perfectly and that they have already evaluated some of the technology available on the market, we set out to gather the customer's requirements and address their concerns.

Initial situation

The client was obviously a startup company and they had not launched their service yet. Initial infrastructure requirements were therefore quite small and simple. What was the real issue was that they had a very broad idea of what their actual load would be so the setup project would consist primarily of stress testing and overloading the infrastructure to set a baseline for the initial requirements and the additional requirements that would incur with growth. Additionally, the customer had some further concerns:

  • Scalability – Would our cloud meet their demands as they grow and would they be able to offer uninterrupted service while scaling their infrastructure.
  • Reliability – The client heard a lot of rumors of some of the cloud providers on the market not offering adequate service levels and numerous downtime reports of some of them prompted the customer to carefully consider this aspect.
  • Price – The pricing of the cloud infrastructure can sometimes be quite high even when compared to traditional environments. Additionally, the calculation can be complex and sometimes even wrong as the underestimating of parameters and their prediction may vary.
  • Support options – The customer offered 24/7 support for their solution but as it would be based on an IaaS cloud, the underlaying infrastructure was expected to offer at least an equal level of support.

Feasibility study

Creating the baseline

Customer's estimation of their application's needs for computer resources was not very accurate, so we set up the initial infrastructure at the maximum of the resources they expected it would consume. Then we started loading the cloud servers with artificial load and carefully observed the graphs that are available through BitSwarm's advanced monitoring. By repeating the process many times in many different scenarios we have successfully narrowed the initial customer's estimate of their needs to a level that seemed reasonable thus creating a theoretical baseline.

Baseline requirements price comparison

We agreed with the customer that the project would consist of two basic parts:

  • Creating the baseline and calculating the price
  • Implementing the solution

This enabled the customer to get help to create a baseline for the infrastructure but not to be limited to a single IaaS cloud provider for actually hosting the solution – The client was able to decide for BitSwarm if it would meet the pricing and other requirements or implement the solution elsewhere if it would not.
With the baseline created the customer calculated the price for the resource rental for at least 5 different IaaS providers.

Addressing the customer's concerns

Price:

The client showed us the calculated prices for us and other competitive providers and our price in that calculation was not the lowest. After reviewing the parameters and realizing the potential long term partnership we have offered the client the following:
  • The client's calculation did not take into account the high price of bandwidth other providers charged. We did a recalculation of the price but with the long term parameters and came up with an actual cost saving on the long run. The client reviewed the recalculation and confirmed it.
  • The long term recalculation also brought larger discounts which further lowered the price.
  • Additionally, our offering of IT professionals in case of server operating system issues added an extra benefit point.

Scalability:

As we have explained the building blocks of our cloud infrastructure to the client, their picture of the possibility of largely scaling their infrastructure was getting clearer. The affirm the commitment we have also signed a custom contract under which we were obliged to assign them more (defined how much) resources in a predefined period of time.

Reliability:

The customer was concerned about our promise of a 100% uptime SLA. They was particularly interested in how have we have managed maintaining it so far. The actual uptime since launching the cloud was:

  • 0.99983% excluding scheduled downtime
  • 0.99966% including scheduled downtime

The customer was satisfied with the resulting uptime but came up with some additional questions regarding how we maintain such uptime. The main points of our answers were:
  • 24/7 on site staff that replaces any failed component
  • all components are hot redundant
  • for each component there is a cold spare part readily available on site

Support options:

Another important issue for the client was how can they contact support and how responsive the support is. We agreed for a testing period in which they would randomly call our support and would measure the response time of our team. The results were satisfactory.

After carefully reviewing our feasibility study and making their own more detailed calculations, the customer decided that they will choose BitSwarm cloud as their IaaS provider. A contract was signed and the project went underway by plan:

  • Create the testing environment for the current production requirements
  • Stress test (by client this time) the testing environment
  • Convert the testing environment to production
  • Create a staging environment for the purpose of testing new versions
  • Create a testing environment on which to try scaling the production environment
  • Evaluate and closely monitor the infrastructure

Conclusion

The project was successfully implemented and after several months of running and scaling the client's infrastructure they confirmed that the solution on partnering up with BitSwarm was the right decision.


↑ Top