In the past few days the failures of the public cloud providers AWS and Microsoft Azure have caused a lot of trouble. At AWS allegedly a typing error during a routine maintenance worked the entire system to collapse. At Azure, users worldwide complained about problems logging in to various Microsoft services. This affected Outlook, Skype and Xbox Live.
Also remain effective in case of failures
At this point the critics of the Hyperscaler could form themselves and the well-known song of the uncertainty and lack of reliability of capacities and services. In principle, however, this case is not primarily concerned with technical questions or the basic justification of concepts. Users need to be aware that the use of globally orchestrated services can lead to global problems. At the same time, cooperation with local service providers has the advantage that problems, if they arise, “remain in the circle of the family”. However, it is relatively clear that problems arise in terms of availability and failure security whenever a single provider or approach is withdrawn. It does not have to be that way today.
Flexible with the Hybrid Cloud platform
The hybrid cloud platform approach of DARZ makes it possible to process the data either in the colocation or the private cloud or in one or more public cloud providers. In the latter case, we are talking about multi cloud sourcing, so it does not mean either AWS or Microsoft Azure. At DARZ we are talking about: AWS, Microsoft Azure, VMWare vCloud Air, ProfitBricks, IBM Bluemix and CloudSigma. And this depending on the need and at different times and for different applications according to customer’s taste. The strict separation between the user’s data header available at DARZ and the computer resources allows a hitherto unknown flexibility in scaling and positioning tasks. In this way, scenarios such as AWS and Azure do not occur at all. Especially because not only the Hyperscaler are available, but also the classic colocation and private cloud variants.
Therefore, our advice is not to think centrally, with the exception of the central dataset, but to be flexible and decentralized in order to remain agile and able to act.