TrueNAS synchronous-write hardware guide
Best SLOG SSD for TrueNAS: PLP, Latency & Endurance
A SLOG is one of the most misunderstood TrueNAS upgrades. It is not a general write cache and it does not accelerate ordinary asynchronous file copies. It provides separate non-volatile storage for the ZFS intent log so synchronous writes can be acknowledged safely with lower latency.
Quick answer
Buy a SLOG only for a real synchronous-write workload
NFS, iSCSI, databases and other sync-heavy applications are the classic candidates. TrueNAS says a SLOG normally needs only enough capacity for a few seconds of incoming synchronous writes, so low latency, high endurance and power-loss protection matter much more than terabytes of capacity.
Live Amazon hardware
Current hardware for this TrueNAS decision
Products come from this sprint's dedicated catalogue. SSD, ECC memory, NIC, DAC and optical classes stay separate. Ambiguous capacities, external drives, USB NICs, enclosures, complete computers, multipacks and incompatible memory types are rejected.
Buying decision
PLP and latency outrank capacity
A huge consumer NVMe with no verified power-loss protection is not automatically a good SLOG. The role exists to protect acknowledged synchronous writes through a crash or power failure. If the exact device cannot preserve those writes, it undermines the reason for using a SLOG.
Interactive decision tool
TrueNAS SLOG Sizing Planner
Use the inputs to narrow the hardware role before shopping. Results are planning guidance. Verify current TrueNAS/OpenZFS behavior, motherboard support, exact SSD specifications, ECC compatibility and backups before deployment.
Compatibility checklist
Four checks before buying TrueNAS acceleration hardware
ARC, L2ARC, SLOG and special are different
RAM is the primary read cache; L2ARC is secondary read cache; SLOG serves synchronous writes; special vdev stores real allocations.
Compatibility comes before price
Verify ECC type, SSD interface/form factor, PCIe lanes, NIC media and motherboard support for the exact part.
Do not infer enterprise features
PLP, TBW/DWPD, ECC mode and module buffering are shown only when the listing or exact specification supports them.
Backups remain separate
ECC, mirrors, cache devices and auxiliary vdevs solve specific failures. They do not replace independent local and offsite backups.
SLOG serves synchronous writes, not all writes
ZFS keeps an intent log for synchronous transactions so it can replay acknowledged work after a crash. A separate log device moves that ZIL activity onto dedicated low-latency storage. Asynchronous writes continue through the normal transaction-group path.
That means an SMB copy or application that does not request synchronous semantics might see little benefit. Identify the protocol and workload before purchasing a log device.
Capacity is smaller than most buyers expect
TrueNAS explains that the SLOG only needs to cover a short window of writes before the normal transaction group commits. It gives roughly 8GB to 32GB as an adequate range for most modern networks, depending on incoming write rate.
Larger drives can still be used because small enterprise SSDs are hard to source, but unused capacity is not automatically performance. Some administrators over-provision larger SSDs to improve endurance rather than using every advertised gigabyte.
Power-loss protection is a functional requirement for the intended safety model
A SLOG lets the system acknowledge a synchronous write after it reaches stable storage. If the device loses acknowledged data during sudden power loss, the guarantee the application asked for has been weakened.
TrueNAS therefore tells users to verify that the SLOG is power-safe or has power-loss/failure protection. The Cloudzat card only shows PLP as verified when listing evidence states it; otherwise the buyer must check the exact vendor specification.
Latency consistency is more important than peak throughput
Synchronous applications wait for acknowledgements, so latency matters directly. A low-latency device that responds consistently under sustained writes can outperform a consumer NVMe whose benchmark strength comes from large queues and burst caches.
Optane-class devices became popular for this reason, while modern enterprise NVMe can also be suitable. Compare latency and endurance data for the exact model, not just interface generation.
Endurance can be concentrated on the SLOG
Sync-heavy systems funnel many log writes through the dedicated device. TrueNAS specifically warns administrators to pay attention to published endurance because the SLOG can become a write focal point.
Estimate daily synchronous write volume and choose comfortable lifetime headroom. Over-provisioning can be useful on supported SSDs, but do not improvise potentially destructive procedures without understanding vendor and TrueNAS guidance.
Mirroring a SLOG is an availability choice
Modern OpenZFS can recover the pool if a SLOG device disappears, but an abrupt loss at the wrong moment can affect the most recent synchronous transactions. Mirroring can reduce exposure to a log-device failure during operation.
The importance of a mirror rises with business impact and replacement time. A home lab may choose one high-quality device; a production database target may justify a mirrored pair.
A UPS complements PLP but is not the same thing
A UPS reduces whole-system power failures and allows orderly shutdown. Device-level PLP addresses what happens inside the SSD when power disappears unexpectedly. Both can be part of a robust storage design.
Do not use “the server is on a UPS” as proof that a consumer SSD behaves like an enterprise PLP device. UPS batteries, cables and power supplies can still fail.
NFS and iSCSI are common SLOG candidates
NFS datastores and iSCSI workloads can issue synchronous writes depending on application and configuration. Virtualization and databases therefore create some of the clearest reasons to investigate a SLOG.
Measure before and after. If the workload is primarily asynchronous or already backed by a fast all-flash pool, a dedicated log can provide little improvement.
Do not force sync=disabled to imitate SLOG performance
Disabling synchronous semantics can make benchmarks look faster by acknowledging data before the safety contract is fulfilled. That changes data-loss behavior rather than solving latency.
If the application requires sync writes, optimize the safe path: adequate pool performance, a suitable SLOG when justified, and tested power protection.
Interface and form factor must fit the server topology
Optane and enterprise SSDs exist in M.2, U.2/U.3 and PCIe card forms. A bargain device is useless if the motherboard lacks the connector, lane allocation or cooling it needs.
Map HBA, GPU, NIC and storage lanes before purchase. A SLOG should not create a new platform bottleneck or force the 10GbE NIC into a slower slot.
Marketplace condition matters for endurance-sensitive devices
Many low-latency enterprise devices are found as used or pulled hardware. Previous writes can matter for a SLOG role, so condition and health history are especially important.
This sprint intentionally focuses its refresh on new listings. A future used-enterprise workflow should expose SMART endurance indicators, seller warranty and health evidence instead of mixing used devices with new ones.
The fastest way to choose is to prove the need first
Measure sync-write latency without a SLOG, identify the incoming synchronous bandwidth, estimate the short log window, then shortlist devices that explicitly satisfy power-loss and endurance requirements.
After installation, rerun the same workload and monitor latency. If the application does not improve, remove the auxiliary log rather than assuming a larger SLOG will fix a different bottleneck.
Questions people ask
TrueNAS SLOG questions
Does SLOG speed up all TrueNAS writes?
No. It primarily affects synchronous writes. Ordinary asynchronous transfers can see little or no benefit.
How large should a SLOG be?
TrueNAS notes that only a few seconds of write traffic must be held, with roughly 8GB to 32GB adequate for many modern networks. Size from actual sync-write rate.
Does SLOG need power-loss protection?
TrueNAS explicitly recommends a power-safe/PLP device because the SLOG protects acknowledged synchronous writes during crashes and power loss.
Is Intel Optane good for SLOG?
Optane-class devices are attractive because of low latency and endurance, but verify the exact model, condition, interface and current availability.
Can I use Samsung 990 PRO as SLOG?
A fast consumer NVMe is not automatically an appropriate SLOG. Verify power-loss behavior and endurance for the exact model; peak benchmark speed is not enough.
Should I mirror the SLOG?
Mirroring can improve log-device fault tolerance and is reasonable where recent synchronous transactions and uptime are important.
Does a UPS remove the need for PLP?
No. A UPS protects the system power path, while device PLP protects data inside the SSD during sudden loss. They address related but different failure scenarios.
Can I remove a SLOG later?
Yes, log vdevs are auxiliary devices and can be removed with the appropriate ZFS procedure. Always verify current TrueNAS/OpenZFS instructions.
Will SLOG help SMB file copies?
Only when the workload is actually issuing synchronous writes. Many ordinary SMB transfers are not limited by the ZIL path.
What is more important for SLOG: capacity or latency?
Once capacity covers the required short write window, latency consistency, PLP and endurance are generally more important than buying a huge device.
Official references and methodology
Verify the exact TrueNAS, OpenZFS and hardware behavior
SLOG recommendations follow current TrueNAS and OpenZFS descriptions of the ZIL/SLOG path. The catalogue accepts low-latency and enterprise candidates but never upgrades an unverified PLP field simply because a model is marketed as enterprise.
As an Amazon Associate, Cloudzat may earn from qualifying purchases. Prices, model revisions, memory compatibility, endurance, PLP, firmware and marketplace conditions can change. Verify the exact part before purchase.