J. Carlos Roldán

Sometimes I have a conversation with an AI that looks particularly interesting and I share it with my friends or colleagues. I thought I could start a new post format, using the typical interview structure, to share it with the rest of the world. This is the first of those. Some replies are shortened (usually by sending "tl;dr" to the AI).

Q: I am running out of space in my computer and phone, and many of my acquaintances pay services like Google One or iCloud. I want to compare alternatives, but first, let's estimate my future storage needs. When I was 12 I had 128GB of storage, by 16 I had 512GB, by 19 I had 1TB. Now I'm 30 and I have 3TB. Plot this in a chart and show me what will be my data needs by the time I'm 45.

A: The projected storage needs by the time you are 45 are around 8.27TB. Here is an exponential plot:

Q: So, let's say I am looking for an storage solution for me, my partner, and possible newcomers. Since I'm a computer scientist, my data needs are greater than the average person. Let's say my partner needs half of what I do, and the new residents of our home will also needs half of what I do in total. What's the total by that time?

A: By the time you are 45, the estimated total storage requirement for your household would be approximately 16.54TB.

Q: Okay, so let's say 20TB to avoid falling short. Is 15 years a realistic measure of time or should I target a shorter/longer period? What alternatives do I have?

A: A 15-year timeframe might be overly optimistic given the pace of technological advancements and increased storage needs, but it is still reasonable. Here are the possible solutions:

• Cloud storage: services like Google One or iCloud for scalable and remote storage options.
• Network-Attached Storage (NAS): a flexible, upgradeable solution that can serve multiple users and functions within a household.
• Direct Attached Storage (DAS): an easy and cost-effective way to expand storage directly attached to your devices.
• Hybrid Solutions: Combine local storage options (NAS/DAS) with cloud storage for a balance of accessibility, security, and redundancy.

Q: I want to find the most cost-effective solution that is reliable to not lose data, allow for individual read/write permissions, and not be too dependent on companies swifting policies (I don't want to have to sign in to Synology NAS or to have Apple double the storage cost suddenly).

A: For a cost-effective and reliable solution with individual read/write permissions that minimizes dependency on company policies:

• Build Your Own NAS: Using open-source software like FreeNAS or OpenMediaVault, you can create a customized NAS without being tied to a specific vendor's ecosystem.
• Use Independent Cloud Storage: Choose a cloud provider with a strong track record of stable pricing and policies, and consider using encryption for added security and privacy.

Q: For reference, I'll paste you here the pricing of a few alternative source storages and you make a table with the 15-year cost. All of them allow at least 12TB storage and at least 4 users, so don't worry about those. One USD is 0.92 EUR, and I want all answers in €. Assume 4 users for the calculations and 50% downloaded per month. [pricings redacted for conciseness]

A: Here's the calculations you mentioned, based on the information provided without accounting for potential changes in rates or special offers over time:

Q: Alright, so now let's do the same estimation with NAS. I want my solution to be reliable, so what RAID options do I have?

A: The following RAID levels have some degree of redundancy:

• RAID 1 (Mirroring): Provides redundancy by duplicating the same data on two or more disks. It's simple and offers excellent fault tolerance, as data can be recovered as long as one drive is operational.
• RAID 5 (Striping with Parity): This level offers a balance of good performance, good fault tolerance, and efficient storage utilization by distributing parity information across all drives. It can survive the failure of one drive.
• RAID 6 (Striping with Double Parity): Similar to RAID 5 but adds an extra layer of fault tolerance by using two sets of parity data. It can survive the failure of two drives, making it more reliable than RAID 5 at the expense of additional storage used for parity.
• RAID 10 (1+0): This naming convention reflects the combination of RAID 1 (mirroring) and RAID 0 (striping). It provides the benefits of both levels: improved performance from striping and redundancy from mirroring. It is very reliable and offers excellent performance but requires at least four drives and results in a 50% loss of storage capacity to redundancy.

Q: Ideally I want not to sacrifice 50% of my storage, and be able to recover automatically from bit rot and from the failure of one drive.

A: Given your requirements—minimizing storage sacrifice, automatic recovery from bit rot, and surviving the failure of one drive—RAID 5 seems to be the most suitable option for your needs.

Q: What disks would I need for a 20TB RAID 5? Do they have to have the same size? Give me the size required for 3-6 drives

A: In a RAID 5 setup for 20TB usable space, all disks should ideally be the same size to optimize storage use and performance. Here's the minimum size required per disk for different total drive counts:

• 3 drives: Each needs to be at least 10TB.
• 4 drives: Each needs to be at least 6.7TB.
• 5 drives: Each needs to be at least 5TB.
• 6 drives: Each needs to be at least 4TB.

Q: I did a bit of research and found NAS enclosures that are not company-dependent for ~200€ for 4 disks and ~270€ for 6 disks. Then I found 10TB drives for 300€, 8TB drives for 220€, 6TB drives for 200€. Compute my options in price. (yes, I'm too lazy to calculate it)

A: Given your price estimations, these are your alternatives with their price:

Q: Alright, so looking at this pricing, it seems that 3-4 drives makes more sense for me. I'd rather go with 3, as it lets me expand the NAS in the future. What do you need to know to include also the electricity bill?

A: To calculate the NAS electricity consumption cost over 15 years, I need to know how many watts (W) are consumed by the NAS, how many hours a day it will be working, and what is the price per kilowatt-hour (kWh) where you are located.

Q: The max energy consumption is 36.5W, my electricity is 0.10689€/kWh, and I want it to run 24/7. Let's assume this price won't change (I hope it scales with my purchasing power).

A: Then the estimated cost to operate the NAS over 15 years, adding a constant consumption, is approximately 513.21€. Note that this computation assumes that the energy consumption is always maximum, which might be higher than the real number.

Q: So finally, update my alternatives table including a row with the cost of running my own NAS. Add a column for a % of difference. Give your conclusion and plot it.

A: Here is the final table. Based on this difference, running your own NAS seems the most viable and cost-efficient alternative.

And that's it. Probably there are many factors that are not considered in this comparison. For instance, I don't need the 20TB right now so I could start with cheaper storage solutions and scale afterward. Also, the cost of fixing possible problems with storage shouldn't be ignored. But the cost of fighting against iCloud to download my pictures to a disk is also higher than I expected. I also want to learn more about what options do I have to see my data besides a file explorer, as I'm sure that the integration won't be as seamless as iCloud syncing with my Photos app.

What do you think, should I have a NAS?