Chia Farm JBOD vs RAID vs ZFS: Which Layout Earns You the Most XCH?

Key Takeaways

• JBOD (Just a Bunch of Disks) is the official recommended storage layout for Chia farming — and for good reason.
• RAID wastes space on redundancy you do not need, because Chia plots can always be recreated after a drive failure.
• RAID 0 is actively dangerous for farmers — one failed drive destroys every plot in the array.
• ZFS is powerful but adds complexity and overhead that most Chia farms simply do not need.^[6]
• MergerFS (Linux) or StableBit DrivePool (Windows) give you a single unified drive view across all your JBOD disks — the best of all worlds.
• RAID 0 across temp SSDs is fine during the plotting phase — just never use it for final plot storage.
• An HBA (Host Bus Adapter) in IT mode is the right tool to connect a JBOD enclosure to your farming rig.

For most Chia farmers, the answer to chia farm JBOD vs RAID is simple: use JBOD. Chia plots are reproducible data, not irreplaceable files, which flips the standard storage logic on its head. This guide breaks down every layout option — JBOD, all RAID levels, and ZFS — so you can build a farm that earns more XCH without wasting money on redundancy you do not need.

What Is JBOD and Why Does It Matter for Chia?

JBOD stands for “Just a Bunch of Disks.”^[7] It is exactly what it sounds like. Each hard drive sits in an enclosure and shows up as its own individual volume on your system. There is no striping, no mirroring, and no parity. Your operating system sees Drive 1, Drive 2, Drive 3, and so on — each completely independent.

In most storage scenarios, that lack of redundancy sounds scary. For Chia, it is actually the ideal setup. The Chia network rewards farmers based on total raw storage capacity. Every terabyte you dedicate to farming is a terabyte that can win a block reward. The moment you introduce RAID, you start shrinking that usable capacity in exchange for protection that Chia farming does not actually require.

The core insight is this: Chia plots are not your family photos. They are cryptographic files that can be recreated from your keys at any time. If a drive dies, you simply swap it out and re-plot. The data is never truly gone — it just needs time and SSD endurance to regenerate. That is a very different calculus than protecting a business database or a media archive.

How a JBOD Enclosure Actually Works

A JBOD enclosure is a chassis built specifically to house a large number of hard drives. It has a SAS expander and backplane, hot-swap bays, fans, and power supplies — but no onboard processor or storage controller of its own. To connect a JBOD enclosure to your farmer or harvester machine, you need a Host Bus Adapter (HBA) card installed in a PCIe slot. The HBA converts that PCIe slot into SAS connections, and a single external SAS cable runs from the HBA to the JBOD, giving your system access to all of the drives inside.

The most important detail: your HBA must be flashed to IT mode (also called initiator target mode or passthrough mode). In IT mode, the HBA passes every drive directly to the operating system as an individual disk. If you accidentally use a RAID-mode HBA, it may try to build RAID arrays in firmware automatically, which is the opposite of what you want. Popular HBA choices in the Chia community include the LSI 9200 and LSI 9300 series, both of which can be found on eBay for $20–50 and are well-supported on Linux and Windows.^[4]

JBOD Enclosure Options for Every Farm Size

Entry-level and hobbyist farmers can get started with desktop cases, USB external enclosures, or small NAS devices.^[5] As your farm grows, dedicated JBOD chassis become the most cost-effective option. The Supermicro SM45 (SuperChassis 847E16-RJBOD1) holds 45 drives in a 4U chassis and is a legendary starting point for mid-level farmers, often available on eBay for $200–400. The Supermicro SM90 (SuperChassis 946ED-R2KJBOD) goes one step further at 90 drives in 4U, accessed through a single external SAS cable. Serious petabyte-scale farmers stack multiple units in server racks, where modern JBODs with high-capacity 20TB+ HDDs can exceed 10 petabytes per rack.^[3]

Understanding RAID: What It Is and Why Most Chia Farms Should Avoid It

RAID stands for Redundant Array of Independent Disks. It is a group of storage technologies designed to combine multiple drives into a single logical unit — usually to add redundancy, increase performance, or both. RAID comes in several levels, each with a different tradeoff between capacity, speed, and fault tolerance.

RAID was built for data you cannot afford to lose. Chia plots are data you can always regenerate — which makes most RAID configurations an expensive mistake for farmers.

Quick Decision Table: Which Storage Layout Is Right for You?

RAID 0: The Worst Possible Choice for Plot Storage

RAID 0 stripes data across multiple drives, meaning every file is split into chunks spread across all disks. The main appeal is speed — read and write throughput scales up with the number of drives. For Chia farming, that speed advantage is completely irrelevant. The Chia harvester performs small, random reads during challenge lookups, and you have 28–30 seconds to respond. A single spinning hard drive is already fast enough. RAID 0 gives you no benefit you can actually use.

What RAID 0 does give you is amplified risk. If any one drive in the array fails, the entire array fails. Every plot on every drive in that stripe set is gone. If you had 10 drives in a RAID 0 array with 9 plots each, you just lost 90 plots in a single disk failure — and you have to replot all of them. With standard JBOD, that same failure costs you only 9 plots on the one dead drive while all 81 plots on the other nine drives keep farming without interruption.

RAID 1, 5, 6, and 10: Better Than RAID 0, Still Not Great for Chia

RAID 1 mirrors your data across two drives, so you always have an identical copy. RAID 5 adds parity across three or more drives, allowing recovery from one failed disk. RAID 6 extends that to survive two simultaneous failures. RAID 10 combines mirroring and striping for both speed and redundancy. All of these are safer than RAID 0 and are excellent choices for storing data that cannot be recreated.

The problem for Chia farmers is the capacity cost. RAID 1 uses 50% of your raw storage for the mirror — so two 16TB drives give you only 16TB usable. RAID 5 with four drives loses one full drive’s worth of capacity to parity. RAID 6 loses two drives. Think about what that means in farming terms: a RAID 5 setup of four 18TB drives gives you 54TB usable instead of 72TB. That missing 18TB could hold roughly 167 additional K-32 plots. Those are 167 chances to win a block reward that you are permanently giving away in exchange for protection you do not need.

ZFS for Chia Farming: Powerful But Mostly Overkill

ZFS is an advanced filesystem and volume manager originally developed by Sun Microsystems. It has features that storage professionals genuinely love: copy-on-write integrity checking, automatic checksumming to catch silent data corruption, native compression, snapshots, and flexible pool management. ZFS can be configured in RAID-Z1 (similar to RAID 5), RAID-Z2 (similar to RAID 6), or simple mirror pools.

For Chia farming, most of these features go to waste. Checksumming and copy-on-write are built to protect data integrity over time — but plot files are cryptographic hashes designed to be recreated, not preserved. Snapshots have no farming value since you never need to roll back a plot directory. RAID-Z1 and RAID-Z2 still carry the same capacity penalty as hardware RAID 5 and RAID 6. And ZFS pool expansion — adding new drives to an existing pool — was a notoriously awkward process for many years, though OpenZFS 2.2 has made improvements.

Jonmichael Hands, VP Storage at Chia Network and the author of the official Chia farming workload analysis, summarized the principle directly in the official documentation: “Plot files contain random cryptographic hashes. If a device containing plots fails, no useful data is lost. Therefore, the best practice for Chia is not to include any data protection.”^[1]

When ZFS Actually Makes Sense for a Chia Farm

There is one scenario where ZFS is a reasonable choice: when you already have a ZFS-based system running and have spare capacity sitting unused. If you are running TrueNAS or an existing ZFS NAS for other purposes and want to add Chia farming to it, configuring each drive as its own ZFS pool (one disk per pool, no RAID-Z) keeps all the JBOD benefits while staying inside your existing infrastructure. Setting up ZFS from scratch just to farm Chia, however, adds complexity without any meaningful reward.

MergerFS: The JBOD Upgrade Most Farmers Are Missing

One of the most common frustrations with pure JBOD is the many drive letters or mount points. On Linux, you might have /mnt/disk1, /mnt/disk2, /mnt/disk3, and so on — each one a separate location where the Chia harvester needs to look for plots. MergerFS solves this elegantly. It is a union filesystem that sits transparently on top of your individual drives and presents all of them as a single directory — for example, /mnt/plots. The Chia harvester sees one folder. Each drive underneath is still completely independent. If disk3 dies, you lose only the plots on disk3, and every other drive keeps working normally.

MergerFS is a Linux tool. Windows farmers have an excellent equivalent called StableBit DrivePool, which provides the same single-folder pooling with a polished graphical interface and no RAID risk. Both tools are considered best-practice by the Chia community and are far superior to actual RAID for long-term plot storage.^[8]

Chia Farm JBOD vs RAID vs ZFS: Full Comparison

The Right Layout for Your Farm Size

Hobbyist Farmers: 1–20 Drives

If you are starting out with a handful of drives, individual disks attached directly to a desktop motherboard via SATA or USB 3.0 enclosures are perfectly fine. There is no need for a dedicated JBOD chassis at this scale. Configure each drive as its own directory in the Chia harvester, or use StableBit DrivePool (Windows) to pool them into one location. Avoid any form of RAID entirely. At this scale a single drive failure might set you back a few days of reploting time, which is far cheaper than losing 20% of your capacity to RAID parity every day you run.

Mid-Level Farmers: 20–100 Drives

This is the range where dedicated JBOD chassis become worth every penny. The Supermicro SM45 is a community favorite: 45 bays, hot-swap capability, staggered spin-up to prevent in-rush current issues, and solid power efficiency at around 75W idle. Pair it with an LSI 9300-8e HBA in IT mode and MergerFS on Linux or StableBit DrivePool on Windows. You now have a single logical farm location, full drive independence, and the ability to hot-swap a failed drive without taking the farm offline. This setup scales cleanly to 90 or 135 drives simply by adding more chassis to the same HBA.

Large and Professional Farmers: 100+ Drives and Petabyte Scale

Petabyte-scale operations move into server racks, multiple JBOD chassis, and potentially multiple harvester nodes connected to a single full node. Used enterprise hardware from eBay — including older SuperMicro, Emu or NetApp JBODs — offers enormous density at very low cost per terabyte. The storage philosophy stays the same: individual drives in IT-mode HBAs, pooled with MergerFS or similar. Some large operations also run monitoring tools like Grafana dashboards on top of Prometheus to track drive health, temperature, and harvester response times at scale.

“The goal of farming efficiently is to store the largest amount of data in the smallest amount of space with the lowest power. Generally, this involves tradeoffs of cost, availability, reliability, cooling, and noise, based on the size and location of the farm.” — Jonmichael Hands, VP Storage, Chia Network^[5]

When RAID Actually Helps Your Chia Operation

There is one narrow use case where RAID earns its keep in a Chia setup: the plotting phase. Creating K-32 plots is a write-intensive process that benefits from high sequential write speeds. If you are using NVMe SSDs or high-endurance SATA SSDs as your temporary plotting storage, striping them into a RAID 0 volume can meaningfully increase your plots-per-day throughput. The key word is temporary. Once the plots are written to your final JBOD storage and the SSD has fulfilled its purpose, the RAID 0 volume is irrelevant. Never use these plotting SSDs as your permanent farm storage.

A second edge case is the very small farmer who has extremely limited plotting capability — perhaps someone relotting on a slow laptop with a single CPU. If a drive fails and reploting would take three to four weeks, the opportunity cost calculation shifts slightly. In that narrow scenario, RAID 5 across a small number of similarly sized drives might be worth the capacity loss to avoid a multi-week farming gap. But this applies to a small minority of farmers, and the official Chia documentation explicitly states that the overwhelming majority of users should not use any form of RAID.^[2]

Real-World Case Study: What Happens When a Drive Fails

Consider two farmers, each running 10 hard drives of 18TB each for a total raw capacity of 180TB. Farmer A uses JBOD with MergerFS. Farmer B uses RAID 0 across all 10 drives for the throughput boost. Drive number 7 fails on both farms simultaneously.

Farmer A loses the plots stored on drive 7 — roughly 18TB worth, or about 167 K-32 plots. The other nine drives keep responding to challenges normally, and total farming capacity drops to 162TB. A new drive arrives in two days, gets reploted over the next few days, and the farm is back to full capacity. Total downtime for the affected capacity: under a week. Total XCH earnings impact: minor.

Farmer B has a different experience. Because RAID 0 stripes data across all 10 drives, every single plot on the array is unrecoverable the moment drive 7 fails. All 180TB — roughly 1,670 K-32 plots — are gone. The farmer must replace the drive and replot the entire farm from scratch. Depending on plotting speed, that could be weeks of reduced earnings. This is not a hypothetical edge case; it is the mathematically guaranteed outcome of RAID 0 when any one disk fails.

Building Your Farm: Practical Setup Checklist

Whether you are setting up your first farm or upgrading an existing one, the fundamentals stay consistent across farm sizes. Start by confirming your HBA supports IT mode and is flashed correctly — the LSI 9200 and 9300 series are reliable options with strong community documentation. Use enterprise-grade HDDs with high capacity ratings; Seagate Exos and WD Gold/HC series drives are popular choices in the Chia community for their density and power efficiency. Avoid consumer-grade CMR drives at scale due to higher failure rates under sustained low-level I/O.

On the software side, Linux farmers should install MergerFS and add an fstab entry that mounts all drives under a single path like /mnt/plots. A typical fstab entry looks like: /mnt/disk* /mnt/plots fuse.mergerfs defaults,allow_other,category.create=mfs,minfreespace=50G 0 0. Windows farmers should install StableBit DrivePool, add all drives to a pool, and point the Chia harvester at the pool’s drive letter. In both cases, the Chia harvester simply sees one plot directory, regardless of how many physical drives sit behind it.

One more critical point: always add new drives to your farm individually over time rather than purchasing a RAID controller upgrade. Mixing drive sizes is effortless in a JBOD setup — you can add an 8TB drive today and a 20TB drive next month without any compatibility issues or array rebuilding headaches.

Conclusion

The chia farm JBOD vs RAID debate has a clear answer: JBOD wins, and it is not close. Chia’s Proof of Space and Time consensus rewards total raw capacity, and plots are reproducible cryptographic data — not irreplaceable files. Every terabyte you lose to RAID parity is a terabyte that will never win you a block reward. JBOD with an HBA in IT mode gives you maximum capacity, simplest recovery, lowest cost, and total freedom to mix drive sizes as your farm grows. Add MergerFS on Linux or StableBit DrivePool on Windows to get a clean, unified plot directory without any of the RAID risks, and you have a setup the Chia community and official documentation both endorse. If you are ready to take your farm’s hardware and software stack even deeper, check out our guide to how Chialisp is transforming blockchain development — because understanding the Chia protocol at a deeper level is what separates hobbyist farmers from serious operators.

Chia Farm JBOD vs RAID FAQs

Is JBOD or RAID better for a Chia farm?

For a Chia farm, JBOD is better than RAID in almost every situation. JBOD uses 100% of your disk space for plots, while RAID sacrifices a portion to redundancy or striping — space that could otherwise be earning XCH.

Why does Chia recommend JBOD over RAID for plot storage?

Chia recommends JBOD because Chia plots can always be recreated after a drive failure, making redundancy unnecessary. The opportunity cost of using RAID — losing usable capacity — almost always outweighs the time it takes to replot a replacement drive.^[2]

Can I use RAID 0 on my Chia farm JBOD setup to speed things up?

RAID 0 should never be used for final plot storage on a chia farm JBOD or any other configuration, because a single drive failure destroys the entire array. RAID 0 is only acceptable on temporary plotting SSDs during the plot creation phase — never on the drives where finished plots live.

What is MergerFS and how does it help Chia farmers?

MergerFS is a Linux tool that combines multiple independent JBOD drives into a single mount point, giving the Chia harvester one folder to scan instead of dozens. It preserves the safety of JBOD — each file still lives entirely on one physical drive — while eliminating the management headache of many separate directories.

Does ZFS work well for storing Chia plots?

ZFS works technically but is generally overkill for Chia plot storage. Its data integrity and snapshot features are wasted on reproducible plot files, and RAID-Z configurations lose the same capacity to parity that hardware RAID does. Use ZFS only if you already have it deployed and have spare pool capacity to dedicate to farming.

Chia Farm JBOD vs RAID Citations

1. Hands, J. (2022). Storage – Chia Farming Workload Analysis. Chia Network Official Documentation. https://docs.chia.net/chia-blockchain/resources/chia-farming-workload/
2. Chia Network. (2024). Farming Considerations: Redundancy & RAID. Chia Network Official Documentation. https://docs.chia.net/reference-client/farming/farming-considerations/
3. Chia Network. (2024). Reference Farming Hardware. Chia Network Official Documentation. https://docs.chia.net/reference-client/farming/reference-farming-hardware/
4. Chia-Network. (2022). Reference Farming Hardware – Chia Blockchain Wiki. GitHub. https://github.com/Chia-Network/chia-blockchain/wiki/Reference-Farming-Hardware
5. Chia Decentral. (2023). Chia Cryptocurrency Farming Guide. https://chiadecentral.com/chia-cryptocurrency-farming-guide/
6. Klennet Software. (2021). What is the Best Storage Layout for a Chia Mining Rig? https://www.klennet.com/notes/2021-05-21-chia-farm-setup.aspx
7. Horizon Technology. (2025). A Guide to JBOD. https://horizontechnology.com/news/a-guide-to-jbod/
8. Perfect Media Server. (2023). MergerFS. https://perfectmediaserver.com/02-tech-stack/mergerfs/