WARNING: VERY LONG POST. I have attempted to clearly define sections with formatting and text cues, but the size of the post and the text font will ultimately result in a pretty big post. Read on if you are a content creator interested in getting the most out of your computer’s expensive CPU/GPU/RAM setup.
Hey guys, I thought with the recent inclusion of build guides that I would take the opportunity to do something a little different and discuss an aspect of workstation builds that often goes untouched: disk setup. To be clear, this guide will focus on workstations that are used for graphical content creation such as photo editing, graphic design, video editing, 3D modeling, VFX, digital compositing, and animation. This guide assumes that the user is a solo or freelance creator, as big-name production companies use entirely different data storage processes (huge high level RAID arrays, servers, and backups upon backups in different physical locations).
With digital content creation, data is everything. Every step of the way, whether it is materials gathering, editing, or rendering/exporting, there is data. What often goes unsaid amidst the flurry of CPU, GPU, and RAM discussions is that the overall disk configuration that one uses as a medium for this data has a large impact on the speed and efficiency of the work being done. Too often will users neglect their disk setup in order to purchase the next best CPU or GPU, while leaving their storage at a simple 2-drive "SSD + HDD" state. However if you skip down to the parts list you will see that this guide recommends a $700 disk setup. Definitely not chump change. This guide is intended for users building workstations in the $1500+ range. The disk setup itself can vary from $300 to over $1000 in cost depending on the configuration choices you make for your budget and/or needs.
So with that said, I will focus on 3 main disk setup goals as they pertain to workstation use: situational speed, efficiency, and redundancy. A basic mid-range disk setup consists of 7 total drives in a 5-volume configuration. The first 3 volumes consist of individual solid state drives (SSDs) used for the bulk of the content creation work. The final 2 volumes are dedicated to data backup and redundancy (hence the 4 hard drives; each volume has a redundant volume).
Much of this guide's efficiency section is a direct application of Harm Millaard's advice for workstation disk setup on the PPBM7 Tweaker's Page. Millaard is a seasoned video editing professional who breaks down what each drive in a given number of drives can be used for and how to choose a drive for editing.
I make a point to call this situational speed rather than simply "speed". There are two kinds of speed to take into account when choosing drives: 1) burst/short read-write, 2) steady state/long read-write. For the sake of the parts in the list, I will limit the discussion of these 2 speeds to SSDs.
If you spend 15 minutes Googling what SSD you should buy, chances are you will hear about such models as the 850 Pro/EVO or Neutron GTX. This is for good reason – they are known to be some of the top or top value performers for users. But what exactly is this “performance”? Performance for what exactly? In most cases these drives are referenced to for their speed in everyday applications. Boot speed, application and game loading, opening files, viewing media, etc. Short periods of load time. This is what I refer to as burst performance, and it has little to do with the bulk of the work one will do for graphical content creation.
Far more important for workstation users is steady state performance. This is the speed of an SSD after enormous amounts of writing or reading data. As you can imagine, being a digital content creator demands such extremes. All SSDs become progressively slower over time, which means that for the drives which are used as either starting points or destinations for rendered or raw data, an SSD that is a better steady state performer is more desirable than one that is a good burst performer.
There are just 3 basic things that one does when working on content: use your program, read your raw data, and write your finished product. Each SSD is devoted to a specific portion of the content creation process and the kinds of speed involved with said portion in order to maximize data read/write efficiency.
Volume #1: OS, Programs, and Pagefile (Samsung 850 EVO 250GB)
- This drive will contain the operating system, all installed programs, and the system pagefile. All of these things benefit from the use of a high burst speed SSD, and that is exactly what the 850 EVO is. It is not the fastest drive on the market for the purpose, but it is as of this posting considered one of the better price:performance propositions.
Volume #2: Raw Media and Project Files (Crucial MX100 512GB)
- Raw videos, photos, or other assets will live on this drive for the duration that you work on a particular project. Project files are those that are created by programs such as Photoshop that reference your work and can be clicked on to open a saved project later (ie, your .psd files). As such, this drive will see a large amount of data movement into and out of it as you work on ever more projects, and for this reason it is important to choose an SSD with high steady state performance. You can find lots of information online as to what drives have better steady state speed. I use TechPowerup’s storage review database. From there, the results of the Crucial MX100 line are very good and combined with the price of the drive present a very attractive option for steady state storage. The relatively large capacity of the drive is to allow for working on multiple projects at a time, all with their own raw materials and project files.
Volume #3: Exports and Preview/Render Files (Crucial MX100 128GB)
- When you want to export your video or hit the render button while editing, your program will create a corresponding file. The location in which this file is created can (usually) be defined by the user. For the same reasons as Volume #2, a steady state SSD was chosen for this task.
Why would one want to divide tasks into specialized drives like this? The simple answer is to avoid any situation in which an SSD has to simultaneously read and write to itself. For example, a video editor with a 1TB SSD has his editing software and his raw video on the same drive and also wants to export his finished work on the same drive. It is easy to see that reading in raw video by the program and writing out the final work all on one drive and data interface is inefficient. By splitting up OS-program duties, raw file reading, and export destination you eliminate any cross reading or writing and allow each drive to devote its full resources to a single task.
You issue commands through your editor via the OS drive, which reads your raw data from the projects drive and spits out a finished result on the export drive.
So you’ve got a fancy set of working drives. That’s great! But what do you do when you’re stopping work for the day or when you’re done with a project? Back. It. Up. In the world of digital content creation your data is the only thing that you have to show for yourself. It is essential to protect your work with data backup and redundancy.
At the minimum this means having one volume for backup and an identical second volume to backup the backup. This is the simplest form of redundancy that can be achieved, because if one drive dies, there is a second copy still available with all of your work on it. Buy a replacement drive, copy the files from the second drive over, and you’re back where you started with safe data. Having just one backup drive means that you lose everything if the drive dies with chances of recovery slim to none and often outlandishly expensive.
As a slightly more focused application of data redundancy, I have chosen to divide data backup into two groups: 1) digital assets 2) raw files, project files, and exports.
Volume #4: Digital Assets (Seagate 2TB main + identical redundant drive)
- This volume will be home to your assets. Assets are materials which you could potentially reuse as pieces of a future project. For example, a logo template, a muzzle flash, a texture, an intro sequence, etc. It is a relatively large 2TB due to the fact that you will want to be able to have several assets in your system easily accessed as you work on new projects. This idea of a long term drive for asset reference is also a reason why your raw files, project files, and exports will not be stored here as they are far larger on a per-file basis than individual assets. Ideally you would want a NAS drive such as the Hitachi Deskstar NAS or WD Red/Red Pro for asset storage, but on a lower budget, Barracudas are decent.
Volume #5: Raw/ Project Files & Exports (WD Red 2.5” 1TB main + identical redundant drive)
This volume will contain all of your raw files, project files, and exports on a per-project basis. You can go with any second volume for this purpose, but for a couple reasons I think the WD Red 2.5” series is pretty much perfect for this task. Here’s why:
As a backup to your life’s work, these drives have the advantage of NAS rating and technology, which decreases the likelihood of data corruption and drive failure.
- The capacity is small enough that the drive is not going to be used for extended periods of time. The less time a drive is used, the less mechanical wear and tear it experiences. As this volume is mainly going to be used for archiving large files of work, it is not as essential to have a specific file on hand for a future project. That is what the assets drive is for. These volumes are meant to be swapped out with fresh replacements as needed. I recommend installing an external or 5.25” hot swap bay such as the IcyDock DuoSwap for the occasions where you would need to reference a source file from a previous project.
- The capacity is large enough that you won’t have 9001 drives laying around.
- The form factor means that the drives will take up less space when out of the computer.
- It is currently the only consumer/prosumer drive that combines all of the above 4 qualities into one product.
Finally, and this may be controversial, but the manner in which redundancy is achieved is NOT via RAID 1 on each pair of drives. RAID 1 gives a false sense of security in that every save made to one drive is mirrored to a second drive. However, every accidental deletion or corrupted file is also made. To avoid this, I prefer to either manually copy files from one drive onto its redundant backup or use a program to mirror one drive to another on a nightly basis. This will help to save you from user error or data corruption.
And that’s pretty much it. This post is already very long, and so I won’t bother with a long conclusion. Hopefully I’ve conveyed the importance of disk setups for workstation usage. It is not something that should be neglected just for CPU/GPU funds, as those are only the engine of the car. Disk setups are both the fuel tank and the fuel line of the car, and perfecting those will allow your system to perform to the max potential of its engine.
A 7-disk setup makes up a significant portion of a build's budget, but you can apply the basic principles to a setup with fewer disks. Perhaps you'll combine the 2 backup volumes into just the Seagates to cut the cost of redundancy in half or maybe you'll replace some of Volumes #1-3 with HDDs instead of SSDs. You can also tweak the capacities of the drives themselves to reach a price point that you are comfortable at. With the budgets that most workstations use it is very doable to have an efficient and safe disk setup. Anything is better than having just 1 SSD and 1 HDD.
Remember, break up and assign your processing tasks to dedicated drives to maximize read/write efficiency, backup your data and backup your backups, and choose the right drive for the job. Thank you for reading!