At the beginning of our project we had little experience in “that cloud thing everyone is talking about”. Just like Gary Klinman from Hewlett Packard pointed out, we already knew that it is possible to “back things up to the cloud”, but in our imagination “the cloud” seemed more like a fancy slogan than a clear technical concept of providing data and processing resources. Sure, as computer science students, most of us already have been developing websites or web applications. Beyond that, we are also using cloud-services ourselves to store our data, although we have learned, not only by the media, that there are securer ways to do so. However we did not know much about the technical concept or architecture behind it. Now, as cloud computing is on the rise and a booming topic in the computer industry, we knew that it would be of advantage to close this lack of knowledge.
So, what is the cloud?
The National Institute of Standards and Technology (NIST) gives the following explanation:
Cloud computing is a model for enabling ubiquitous, convenient, on-demand network access to a shared pool of configurable computing resources (e.g., networks, servers, storage, applications and services) that can be rapidly provisioned and released with minimal management effort or service provider interaction.
Therefore the NIST specifies the following five characteristics of cloud computing:
- On-Demand Self Service: The consumer can manage his IT-capacities whenever he needs to, without any interaction with the service provider. It is also possible to adjust them automatically.
- Broad Network Access: Thanks to standardized network access, users can connect to the cloud with any device (laptops, mobile phones, tablets, etc.).
- Resource Pooling: Computing resources of the provider have to be pooled so they can be served and assigned to multiple consumers. As a result the consumer has access to all of his resources without the need to handle them himself.
- Rapid Elasticity: Capabilities can be scaled up or down elastically and fast, sometimes even automatically.
- Measured Service: Due to automatically resource optimization and metering the consumer does never pay for more resources than he actually needs.
So much for the theory! With this knowledge in mind it should have been like stealing candy from a baby to build our own cloud, when we realised that we would not have the budget to set up our own cloud server hardware. For this reason we had to choose a service provider, which we finally found in IBM’s cloud platform Bluemix, about which we will report more in the next blog post. Nevertheless we still had to design and implement our cloud application’s architecture ourselves.
Dark clouds on the horizon – or challenges?!
As it turned out it was not that easy to set up our cloud application and so we encountered a bunch of problems. It soon became pretty clear that we have to think in bigger dimensions and regarding some aspects we did not pay much attention while setting up the application in the first place in a local environment. For example we had to adjust our architecture several times to avoid bottlenecks, which would have slowed down or stopped the entire process, due to a lack of capacity to handle the occuring volume or traffic. We also learned that we should not be stingy with resources, therefore (and for maintenance reasons) we decided to assign one database for each social network. Additionally, we had to handle some unexpected impediments like an incorrect or outdated documentation or a temporary unavailability of Bluemix lasting for hours. It can only be hoped that the reason for the latter is that we used a free trial-account and that Bluemix works stable for paying customers.
Well, in the end we were able to reveal the cloud’s secrets and we hope this post cleared up the cloudy mystery at least a little bit and maybe even brightened up your day! Make sure to catch the next blogpost when we will be looking at cloud service models and the earlier mentioned cloud platform Bluemix.
Written by Natali Bopp, Andreas Gold, Jonas Häfele, Merle Hiort, Martin Kopp, Christian Lang, Anna Poth und Eric Weislogel
Leave a Reply
You must be logged in to post a comment.