If you can afford a Bugatti Veyron, then you can afford a data center (makes you wonder what all these Veyron drivers are doing with their lives... I'm talking to you Ronaldo!). You know you're sitting with people who play in the big leagues when the economics discussed sound like pro soccer salaries (data center infrastructure is no game for those of us still opting for the penny machines in Vegas). The exact dollar value estimated for starting a data center can exceed millions, but the hyper converged trend has the potential to cut that figure down to a modest Rolls Royce. Of course, the numbers vary in both directions depending on the size of the data center, and the bottom line is only one factor that makes hyper converged infrastructure a trend worth exploring for corporations.
Now that you past the test and avoided becoming distracted by expensive cars and good looking athletes, let's talk about what hyper converged actually means and another important question, "why?". Hyper-convergence aims to combine three key components of data centers: computing, networking, and storage. Traditionally, such components (like VM computers, data storage devices, servers, switches, routers, etc.) exist separately. An IT person would need to become familiar with how all these components interface with one another physically and through software in order to mitigate risks (because problems between interfaces create the risk system failures). In a converged system, multiple components are engineered and pre-configured to reduce compatibility issues, but the components could still be separated if necessary. In a hyper-converged situation, components are tightly integrated and designed to work as one (they can not be separated). Thus, convergence offers a "package solution" to the problem of incompatible hardware/software, buying multiple pieces of hardware, and having to troubleshoot multiple interfaces. It can also improve computing efficiency (by measurements as high as 73%), and save energy (my favourite part! Which also happens to save money on power).
At the IT Pro meeting, the speaker specifically discussed a hyper-converged infrastructure solution provided by Gridstore. Gridstore offers an all-flash complete system that includes compute and storage nodes, uses Windows Servers and special software, and is highly scale-able and affordable for "smaller" corporations (as compared to "bigger" ones like Google). They also solve an issue with predictable performance in virtual environments. This product is created by an alliance of tech giants. Gridstore's partners include big names in tech like Microsoft, Intel, Dell, Vision Solutions, and Citrix (check it out here). Hence, the system has ample support. I have witnessed the in-depth customer support that comes with enterprise level Microsoft products in both work environments and when I attended the meeting (I think they deserve a shout out for that).
The next speaker, Dave Kawula, a Microsoft MVP, joined the meeting remotely to discuss some of the software used in the system that greatly improves the speed and user friendliness of virtual networking and node configuration. At this point the conversation became a lot more complex with topics like parallel flash architecture and more virtual computing details (I definitely need to brush up on my parallel flash architecture and my back-end switch plane knowledge...:-/). To redeem this paragraph of the blog post, I will leave you with a nice little list of topics that all relate to this:
In summary, I followed up my data center adventure with a glimpse into the emerging trend of hyper-converged data center infrastructure. I gained new insight into the potential risks (and number) of problems between hardware and software interfaces, and how converged technology can mitigate them by integrating computing, networking, and storage. The meeting also introduced me to the vastness and complexity of innovative solutions like the All-Flash Hyper-Converged System developed by Gridstore and I gained an appreciation of the economics involved in running and scaling a data center, and its relation to infrastructure. Plus I got a free sandwich. All in all, a good day.
No comments:
Post a Comment