To upgrade or not to upgrade, that is the question



Ever since I got my first PC at 12 years old and wrote my first DOS & Pascal programs (yes that was before Windows), I have always been interested in technology and how to harness its power.


My school at the time had 4 classes of the same grade with about 40 pupils per class. Rosters were printed and supervision classes (when teacher's were absent, you'd go to a randomly assigned supervision class) dynamically assigned at the beginning of the school year. The problem with this system was that it would split up friends, which of course was no fun if you are 12 years old. My first "project" was to change my roster to include all the supervision classes I wanted to be in. I also noticed that there were patterns as to when I would have Science and my friends Math and by simply swopping these around I could be in their classes too. Word spread and soon I was printing rosters for students at R5 (roughly $0.50) each. Teachers didn't know what was happening and as the automated rostering system brand new and ground breaking for the school, they figured it must be a system issue.


Throughout my career, from Government customers to Telecoms to Enterprise and SME's to my own business, the mission has always been the same - to maximise return on investment - irrespective of whether its time or capital. This philosophy made me quite successful over the years, because quite simply who doesn't want to get more for less?


As I'm writing this post on my Desktop PC, which is approaching it's 5th anniversary, the question that comes to mind is when is too old, too old?


Corporate standards dictate that a refresh is needed every 2 to 3 years while Small to Medium businesses will generally sweat its hardware till its last breath (or spark I should say!). While a refresh every 2 to 3 years will keep a hardware estate in reasonably fit for purpose condition, it does mean that hardware spend is quite a bit more than what is really needed. SME's on the other hand of the scale will spend far too much time and money on supporting old hardware than what its worth. So where is the sweet spot?


The answer, seemingly simple, has a lot of different variables:


Form Factor

Does your machine need to be portable or not? Laptops are far more expensive and generally less upgradable than desktops. I once worked for a software company where everybody had laptops, but nobody ever travelled with them. They literally stayed, plugged in, in the office. So in reality, they spent 50% more per full time employee for hardware than they needed to. Combine that with their 2 year refresh cycle and that's a lot of wasted funds that could have been directed elsewhere.


Use case

Lets consider, most importantly, what the machine is going to be used for? Is it purely a machine to access the internet? Does it only need to run general Office applications or does the business use specific applications which require very specific resources? In the case of the latter, is there a roadmap available for next releases and expected system requirements?


Lifecycle

Within the SME space, lifecycle is often overlooked and many businesses only look at what they need for their work today. It is important to map out how long the machine is expected to last, not only in respect of hardware, but also software requirements. Also, at the end of its lifecycle, can it be repurposed in another role?


Expandability

This ties in with the lifecycle strategy and a good machine should be able to have different components upgraded as needed. Common components such as memory, storage, CPU and Graphics should be easily accessible.


Budget

Of course, just like most things, you get what you pay for. I have had clients that purchased laptops on "special" from large chain stores only to complain that it is...slow. This is where we can easily quantify what that cheap purchase is actually costing your business. It's not to say that all machines from chain stores are bad, just that the buying decision was made, based on the incorrect variable.


Brand

The battle between INTEL and AMD is almost as old as PC's themselves with these two rivals starting operations just shy of 1 year apart, more than 50 years ago, both in California, in the good US of A. Personally I have always been an AMD fan, simply because they were cheaper for the same performance as Intel. Intel's brand however promotes compatibility and stability - which it does very well. Since AMD launched its Threadripper series of processors utilising 7nm*, it has consistently outperformed Intel's best chips produced at 10nm.

nm refers to the density of the silicon layers in the chip - the smaller the number, the closer together they are, which in turns means more can fit which in turn means a faster chip.


So back to my Desktop, which still has a good 2+ years left without any upgrading I would suspect. I got my current PC as an INTEL I5, 16GB Memory in 2016. My desktop is used for work (a myriad of applications, but nothing too CPU intensive) and play (I'm an avid gamer so good graphics and sometimes CPU is important too). Since I have purchased my PC, I have upgraded the CPU, memory, storage and graphics, thereby making my machine last a good 7 years. These were not all upgraded at the same time and by targeting specific components at specific times I always got the maximum performance for what I needed vs cost at that time. I can play AAA titles with my current GTX1070 at 70-100 FPS, but the next component would be the graphics card again as a lot has changed in the past 2 years in the world of graphics.


So what did I do to achieve this?

Firstly I made sure that the motherboard that I chose had all the ports I needed, USB 3.1 was the newest standard out at the time - today we have USB4.0. USB 3.1 in dual channel and USB 4.0 both have the same throughput at 20Gbps.


Secondly I looked at CPU socket support. LGA 1151 was relatively new in 2016, meaning that today I can still purchase a brand new 9th Gen INTEL CPU that will work quite happily on my 5 year-old motherboard.


I then looked at max memory slots and support. My initial 16GB memory was based on 2 x 8GB 3200Mhz chips. I quite easily upgraded these to 32GB by adding another 2 x 8GB with the same memory profiles. My board supports up to 64GB which is far more than I would need before replacing the board itself.


Network is also very important and I ensured that my board came with at least 1xGbps network port. This has been the gold standard for many years, though some of the newer high end boards will have 10Gbps integrated. Of course for roughly $150 a dedicated 10Gbps network card can be added without any issues.


SLI is supported, meaning that I could add another graphics card to improve graphics performance without replacing the original card.


Storage was initially on 1x1TB SSD, which was quite pricy 5 years ago, however the performance boost over normal hard disk drives warranted the additional expense. Today this has been expanded to include and additional 3TB of SSD storage.

I did include a DVD drive, primarily for being able to write to DVDs, which you've guessed it...I've never used - not even once!


So where to from here?

As stated before, next I could upgrade my graphics card. Even though my board is Gen3, and the latest GTX3080 is Gen4, it will sit nicely with negligible performance difference.


Conclusion

By taking all the necessary variables into account and mapping a firm lifecycle for each piece of hardware, businesses are able to reduce their IT overhead substantially.

As can be seen in the above example, the lifecycle does not necessarily dictate an end, but rather a continuous roadmap for improvement to realise the best possible ROI.