Motorcycle riding is my passion. I got my first minibike, a two-wheeled contraption powered by a lawn mower engine, when I was eleven years old, and I’ve been riding ever since. I love the ride itself, the sights and the smells. Equally important, the ever-evolving technology is pretty cool. I am always adding enhancements to my bike, to give it more power, comfort, capabilities, storage and speed. I cannot leave a stock motorcycle “stock”. Personalization is the way to go.
I have the same passion for IBM Spectrum Scale, high performance scale-out storage software and enterprise data services with global federated capacity, as well as Spectrum Scale RAID, simple scalable building blocks for high-performance IBM Spectrum Scale hybrid cloud storage solutions. I enjoy working with our clients to get the most out of their Spectrum Scale implementations, helping them to reach their business objectives.
I’ve written about this comparison between motorcycle riding and IBM Spectrum Scale before, explaining that scaling up and scaling out a motorcycle is much like Spectrum Scale capabilities. You can find that blog post here.
The hardware that IBM Spectrum Scale runs on can be thought of as the engine of a motorcycle. If you want to increase the power of your engine, you scale up Spectrum Scale by upgrading to a more powerful system, adding more processing power, memory and so forth. The more horsepower a motorcycle has, the less effort is required to move me from one location to another. In the same manner, scaling up Spectrum Scale will allow it to operate faster and more efficiently.
The “scale out” process is akin to adding storage capacity to your motorcycle. Spectrum Scale can be scaled out by adding additional components — in this case, more disk or solid-state drive (SSD) storage. We can further scale out Spectrum Scale by adding additional NSD servers to the cluster. This would be similar to adding an additional motor to a motorcycle or replacing the two-cylinder motor with a four-cylinder motor. However, in Spectrum Scale, the process of scaling out is much easier!
Today, this analogy still feels like a good fit.
If you’re talking about motorcycles, you can buy a new model with new features. The new bike would have features such as “rear cylinder shutdown” so it does not overheat in traffic. This is applicable to air-cooled engines only. Another feature is software that controls the performance of the engine. For instance, many motorcycles now have a “regular, sport and cruising” selection which changes the handling, air and fuel ratios depending on which mode is selected. If you are into the gadgets, you can upgrade, or like me, you can keep your older motorcycle, and take advantage of the software updates that have new features. This would enable you to have some (GPS updates, satellite radio, ride mapping) features that may not be part of your original motorcycle. This will vary, because just like software and hardware, there are limitations to upgrades.
As for Spectrum Scale and Elastic Storage Server (Spectrum Scale GPFS Native RAID), new models and software have eclipsed the old models. Some of the software is no longer supported, and the base needs to be upgraded to support the new change. But there’s always the option of remaining on the current configuration until such time as you must upgrade/update or run the risk of not being supported. I don’t recommend that. It is better to upgrade before the end of life of a product/solution.
It’s always a matter of decisions, decisions. Do I purchase a new motorcycle? Or do I stay where I am and upgrade the software that manages and powers the bike? Spectrum customers have to make similar decisions when they think about upgrading the software that currently runs their Spectrum Scale, ESS or any other infrastructure in their data centers.
Software updates can, and do, fix issues, add features and functions, and further enhance the operation of the systems. It is something that should be considered, especially if you are N-2 or more releases behind. It is highly likely that that at an N-3 level, your software stack is no longer supported and upgrading is a necessity. Look at it this way, you have built a very reliable, comfortable motorcycle with a great rider experience (think GPS, music, comfort) that supports your “workload” reliably, and performs very well. The newer upgrades to the software and even hardware (engine, transmission) will further enhance your ride. Why not do the same for your software-defined infrastructure (SDI) and software-defined storage (SDS) infrastructure?
While motorcycles are perfectly fine and supported for decades, it is not the same with business-critical operations that have a direct impact on your customers. You must provide the most up to date, fastest, reliable SDI/SDS infrastructure. If you do not, then you risk customer satisfaction issues, and even more important, you risk impacting your business.
Customers need to look at all the possible changes. Some are critical. But others, need to be examined closely. How will they positively or negatively impact the business? “Who is this change going to impact”? What is the impact to “the Who“? And “Why” should I change? If you can answer the “Who, What and Why” question with more pros than cons, then it could be very beneficial to upgrade for more “scale out and scale up” capabilities.
IBM Systems Lab Services Storage is a team of consultants with Infrastructure expertise to help you build the foundation for today’s hybrid cloud and enterprise IT data centers. Contact us today if you’re looking for support to scale up and scale out your storage infrastructure.