Servers are the batteries of the cloud
The 18650 battery is used in almost every device you can think of. You put 1 battery in a flashlight, 2 in a vape pen, 6 in a laptop, 8 in a drill, 100 in an ebike and 1000 in a car (yes, even Tesla just use these commodity type batteries in their cars).
Batteries are commodity, heterogeneous and easily replaceable - basically you scale up the hardware depending on the application. In a lot of ways server hardware in the datacentre has become the 18650 battery of today for the cloud.
It powers every app we use, but nobody really notices it.
Back in the early 2000s hardware and software were intrinsically linked in peoples minds. . Fast forward 20 years and hardware of course still forms the fundamental basis of the technology pyramid, however, it has become abstracted, somewhat benine even from the perspective of IT administrators worlds. Servers live in a datacentre, somewhere off in datacentreland, or in a cupboard nobody wants to go to if you are an SMB (unless it has air con and it is a hot day!). Servers are usually clustered and heterogeneous these days.
Like the batteries of a device - most people don't see server hardware, or particularly care about it - unless it stops working, then it is a problem!!!
I feel much like the user experience has become the key to successful apps, it is now the key to successful virtualization and hyperconverged software/hardware platforms.
No longer do administrators want to be encumbered with navigating linux clis or windows shells, far less server BIOS and IPMI interfaces. Modern software products now give you the ability to deploy physical server hardware much like the venerable battery we looked at in the beginning - without a great deal of customization or thought. Through these changes and the increasing adoption of off-site datacenters (aka cloud), hardware has become commodity. If you want to make something big - you deploy lots of servers under the software defined layer of your choice. Usually Microsoft, Nutanix or VMWare. You deploy with your right mouse finger, not by opening a box of bits and stuffing it in a rack!
However, overlooking the hardware aspect of any server deployment would be foolish! Not only can the underlying hardware selected have a huge bearing on TCO which ultimately forms the basis of your ROI calculations (or should), it can massively affect application performance in the process. Simply selecting the wrong CPU could mean you deploy 2x the amount of servers your application needs to perform effectively. Not considering new technologies like NVMe or Hyperconverged, could put your company 3 years behind the competition in terms of IT efficiency. Purchasing servers off the shelf vs bespoke might mean you are paying a 30% premium for un-needed features. The list goes on and these are not things to be taken lightly - they are things that can mean success or failure of a business model.
The amount of times I see people take minimal care over their hardware BOM is significant - statements like "Just any 8 Core server with 128Gb of RAM" end up filtering down to my teams. This sort of situation is where a solution architect at Tech Data has the opportunity to step-up, and ask a number of questions to better understand what the customer is looking to do and how their application would benefit from more analysis and thought taken on the hardware BOM.
Example BOM summary for an Edge/ROBO Hyperconverged Infrastructure Solution suited to SMB. Retains central platform commonality across multiple variations of BOM thus decreasing support cost and increasing flexibility.
At Tech Data we de-mystify this often overlooked hardware aspect on behalf of hundreds of customers every day, acting as the trusted advisor for everyone from an SMB looking to deploy an on premise hybrid cloud storage solution, to an ISV specifying hardware for an appliance, or an MSP looking to deploy hardware in the cloud for resale. Our skilled architects and vendor agnostic approach ensures you are presented with the best BOM for your solution. We also ensure that the solution is matched to the software requirement - not matched to that months excess server inventory! ;)
Contact me to to find out how to engage with our solution architect team via private message or email.
"The amount of times I see people take minimal care over their hardware BOM is significant" - I couldn't agree more! Those who don't take the time to understand the impact of hardware choice on their solution stack, are often leaving performance and value on the table. For an example - look at the below test result shared at a recent Intel Memory & Storage Day, where a 3rd party test showed 61% more transactions per minute achieved simply by swapping out a SAS SSD cache drive, for a Intel Optane SSD DC P4800X based cache. Never say hardware doesn't matter! https://newsroom.intel.com/wp-content/uploads/sites/11/2019/09/Intel-2019_MemoryStorageDay_FrankOber_final.pdf
What a cool title! Love it.