I've been meaning to put something like this together for a while. In recent weeks I've found myself writing this more and more so thought I would just post my thoughts in a thread that I could link to.
So the reason for this thread is to try and give insight as to the reasoning behind having more than one drive in a build, typically one for the OS and one for the games/fast storage. There are often more than 2 drives recommended, but there are varying reasons for this as well. The main consideration here will be why I would never recommend having everything on one drive.
The short version looks like this:
The longer version looks like this:
Due to the nature of SSD drives, the fuller they get.... the slower they get. This is typical of all drive types, but with SSDs it's far more noticeable. Different manufacturers have different thresholds on when this performance starts to take a hit. 50-70% is the typical values. Around the 50-70% mark the difference would be noticeable with the OS performance tasks, not so much with gaming/regular storage. After the 70/80% mark this will become noticeable with even day to day tasks. After 90% the performance tanks and falls off a cliff. Having a reduced performance on a non-OS drive will never be anything more than a slight inconvenience, but having it on a primary drive will lead to significant annoyance IMO. I believe most would happily spend the small difference to have this quality of life guarantee.
With the above in mind, and factoring in that keeping the OS running well is the most important consideration here, any drive storage would need reduced to 50% utilisation. If you were to purchase a 1TB drive, this would be limited to 500GB of use, with the OS/pagefile/etc taking up around 80GB, and a typical program package in the region of 20-40GB, I would typically consider 120GB being "Windows". This would leave 380GB for storage use, which isn't a lot nowadays. If you were to upgrade this to 2TB, you would have an increase to 880GB. This would be far more typical of usage nowadays. A 2TB drive will typically cost more than a 500GB drive and 1TB drive together. With the 500/1000 solution in place you could easily have 120GB put to "Windows" a further 120GB on the primary drive for fast storage or the odd game, with the secondary drive easily able to aim for 80% capacity without any real down-sides. This would allow for over 900GB of storage on this solution, it would typically be cheaper as well.
More drives = more performance. With the same reasoning as why RAID can be setup to run faster by utilising dual "lanes", the drives by design run faster when separate. Each drive is connected to it's own PCIe lane interface. This, certainly in most cases, is a direct link to the CPU via the motherboard. When reading/writing data, it is sent via the lane to be processed by the CPU. Each drive sends data and it queues waiting on the CPU to process. If you think of the CPU as a rotating clock (you'll note the reference here) each time the clock hand passes the drive lane, it does something. If there are 100 processes queued, it will take 100 passes. This affects performance by the makeup of those queued processes. If they are all OS based, the OS will be running optimally, if they are a mix of OS tasks and game loading tasks, the queue wait time performance will be split between the 2 operations, before even getting to the CPU. If you have 2 drives doing 2 queue operations, each task will be handled in parallel, with the CPU processing them both during the clock cycle. This is a very rudimental description to try and paint a picture of how it works.
Single point of failure. If you are in any industry with modern controls in place this will be a very familiar phrase to see. Having everything in one drive is not a good idea. Even with backing up, you're leaving yourself open to a management headache if there were to be a failure. If the OS drive fails, it's a case of installing the OS (fairly quick nowadays). If the game drive fails, this is more problematic as the downloads can take an age.... but at least you can attempt to get things in order and have a functioning system during the headache.
In a less catastrophic failure situation where Windows needs installed it's always better to wipe the drive partitions and start from fresh. Nowadays Windows uses multiple partitions for security and boot strategy. When installing from fresh you want to completely clear everything to remove any boot records or partitions and the easiest way to do this is by blanketing the drive. If you had programs/storage on the drive being used you would need to be very careful and it's always far more difficult to ensure that everything has been removed appropriately. This would even be recommended with major Windows updates, and even just as an exercise to know what to do if the drive fails etc. It's a fundamental skill to have with any sort of PC ownership IMO. Partitioning can help avoid this sort of thing, but again you would need to know your way around the partitions and the performance hit from partitioning an SSD just isn't worth it (you eat into storage limitations and queue limitations as discussed earlier, as well as read/write location juggling).
If you have everything bespoke/customised (games, documents, etc) in a secondary drive and you are looking to upgrade to a new system..... you simply take your drives over intact with all the data. You can take over your primary drive as well, but Windows would need a fresh install to take care of the new hardware in play and to reduce any conflict issues. If you had everything on one drive, sure you can carry that across but the Windows installation will include the same considerations as the drive failure situation above. With everything on the one drive this would be quite perilous if you weren't 100% in how this all works.
Following all the guidance leads to good storage habits as well. This should never be overlooked. Sure, it's a minor advantage but knowing where everything is and having a place for everything is a fantastic benefit in any modern system. I tear my hair out looking at friends and family's systems at times, purely because of how disorganised they are. Almost every time I look at something, the immediate fix is a format as they've filled the primary full of faff that isn't required. They have no idea that most of it is even there as everything they actually want is backed up (thankfully they listen to me here).
In conclusion, there's no down side. Even financially there's not enough of a difference in price to be considered saving overall. The cost of having 2TB of actual storage isn't simply a 2TB drive, you would need 4TB..... not because of the 50% rule, but simply due to the way the M2 drives are designed.... there's no such thing as 3TB.
Hopefully this helps anyone trying to understand the reasoning. Whether you're happy with the quick hit list at the top or the more in-depth view, I hope it gives confidence that its a considered logic that's taken when I offer the advice. It's not simply "what I do" that I think is best, it's the reasons why I do it that I think it's best.
So the reason for this thread is to try and give insight as to the reasoning behind having more than one drive in a build, typically one for the OS and one for the games/fast storage. There are often more than 2 drives recommended, but there are varying reasons for this as well. The main consideration here will be why I would never recommend having everything on one drive.
The short version looks like this:
- The more full a drive, the slower it operates. Modern drive performance reduces significantly after 60% utilisation.
- If you keep it performing well for the OS with 60% utilisation, you lose 40% capacity.
- Each drive has it's own connection to the CPU/RAM (called lanes). More lanes = more performance as you can do something on each drive simultaneously.
- You will have all your eggs in one basket. If it fails, you lose everything.
- If you need to clean install Windows, you don't need to worry about backing up a huge amount of data, everything is already neatly stored.
- Transferring your game library/storage to a new build is as simple as transferring the drive itself.
- You're coerced into following good storage habits, this is great for housekeeping & file structure management.
The longer version looks like this:
Due to the nature of SSD drives, the fuller they get.... the slower they get. This is typical of all drive types, but with SSDs it's far more noticeable. Different manufacturers have different thresholds on when this performance starts to take a hit. 50-70% is the typical values. Around the 50-70% mark the difference would be noticeable with the OS performance tasks, not so much with gaming/regular storage. After the 70/80% mark this will become noticeable with even day to day tasks. After 90% the performance tanks and falls off a cliff. Having a reduced performance on a non-OS drive will never be anything more than a slight inconvenience, but having it on a primary drive will lead to significant annoyance IMO. I believe most would happily spend the small difference to have this quality of life guarantee.
With the above in mind, and factoring in that keeping the OS running well is the most important consideration here, any drive storage would need reduced to 50% utilisation. If you were to purchase a 1TB drive, this would be limited to 500GB of use, with the OS/pagefile/etc taking up around 80GB, and a typical program package in the region of 20-40GB, I would typically consider 120GB being "Windows". This would leave 380GB for storage use, which isn't a lot nowadays. If you were to upgrade this to 2TB, you would have an increase to 880GB. This would be far more typical of usage nowadays. A 2TB drive will typically cost more than a 500GB drive and 1TB drive together. With the 500/1000 solution in place you could easily have 120GB put to "Windows" a further 120GB on the primary drive for fast storage or the odd game, with the secondary drive easily able to aim for 80% capacity without any real down-sides. This would allow for over 900GB of storage on this solution, it would typically be cheaper as well.
More drives = more performance. With the same reasoning as why RAID can be setup to run faster by utilising dual "lanes", the drives by design run faster when separate. Each drive is connected to it's own PCIe lane interface. This, certainly in most cases, is a direct link to the CPU via the motherboard. When reading/writing data, it is sent via the lane to be processed by the CPU. Each drive sends data and it queues waiting on the CPU to process. If you think of the CPU as a rotating clock (you'll note the reference here) each time the clock hand passes the drive lane, it does something. If there are 100 processes queued, it will take 100 passes. This affects performance by the makeup of those queued processes. If they are all OS based, the OS will be running optimally, if they are a mix of OS tasks and game loading tasks, the queue wait time performance will be split between the 2 operations, before even getting to the CPU. If you have 2 drives doing 2 queue operations, each task will be handled in parallel, with the CPU processing them both during the clock cycle. This is a very rudimental description to try and paint a picture of how it works.
Single point of failure. If you are in any industry with modern controls in place this will be a very familiar phrase to see. Having everything in one drive is not a good idea. Even with backing up, you're leaving yourself open to a management headache if there were to be a failure. If the OS drive fails, it's a case of installing the OS (fairly quick nowadays). If the game drive fails, this is more problematic as the downloads can take an age.... but at least you can attempt to get things in order and have a functioning system during the headache.
In a less catastrophic failure situation where Windows needs installed it's always better to wipe the drive partitions and start from fresh. Nowadays Windows uses multiple partitions for security and boot strategy. When installing from fresh you want to completely clear everything to remove any boot records or partitions and the easiest way to do this is by blanketing the drive. If you had programs/storage on the drive being used you would need to be very careful and it's always far more difficult to ensure that everything has been removed appropriately. This would even be recommended with major Windows updates, and even just as an exercise to know what to do if the drive fails etc. It's a fundamental skill to have with any sort of PC ownership IMO. Partitioning can help avoid this sort of thing, but again you would need to know your way around the partitions and the performance hit from partitioning an SSD just isn't worth it (you eat into storage limitations and queue limitations as discussed earlier, as well as read/write location juggling).
If you have everything bespoke/customised (games, documents, etc) in a secondary drive and you are looking to upgrade to a new system..... you simply take your drives over intact with all the data. You can take over your primary drive as well, but Windows would need a fresh install to take care of the new hardware in play and to reduce any conflict issues. If you had everything on one drive, sure you can carry that across but the Windows installation will include the same considerations as the drive failure situation above. With everything on the one drive this would be quite perilous if you weren't 100% in how this all works.
Following all the guidance leads to good storage habits as well. This should never be overlooked. Sure, it's a minor advantage but knowing where everything is and having a place for everything is a fantastic benefit in any modern system. I tear my hair out looking at friends and family's systems at times, purely because of how disorganised they are. Almost every time I look at something, the immediate fix is a format as they've filled the primary full of faff that isn't required. They have no idea that most of it is even there as everything they actually want is backed up (thankfully they listen to me here).
In conclusion, there's no down side. Even financially there's not enough of a difference in price to be considered saving overall. The cost of having 2TB of actual storage isn't simply a 2TB drive, you would need 4TB..... not because of the 50% rule, but simply due to the way the M2 drives are designed.... there's no such thing as 3TB.
Hopefully this helps anyone trying to understand the reasoning. Whether you're happy with the quick hit list at the top or the more in-depth view, I hope it gives confidence that its a considered logic that's taken when I offer the advice. It's not simply "what I do" that I think is best, it's the reasons why I do it that I think it's best.
