Dell PERC H730P: Cache Battery, RAID Mode and Migration
The PERC H730P is a Dell 12Gb/s SAS RAID controller with battery-backed cache, fitted across a wide range of PowerEdge servers. There is one thing about it that fails silently and costs more performance than any configuration choice.
The cache battery is the silent failure
Worth putting first, because it explains a great many complaints about slow servers.
The controller carries cache memory that absorbs writes and acknowledges them immediately, which is where a large part of its performance comes from. That is write-back mode.
Write-back is only safe while the battery or capacitor module can preserve cache contents through a power loss. So when the battery fails, the controller switches to write-through — every write goes straight to disk and is acknowledged only when the disk confirms it.
The consequence: write performance collapses, and nothing appears broken. No failed drive, no array error, no obvious fault. The server is simply slow, and it stays slow until somebody checks the controller status.
Two practical points. Cache batteries are consumable with a limited life, so this is a scheduled replacement rather than an unexpected failure. And a controller relearning its battery temporarily drops to write-through too, which explains periodic slowness on a schedule.
Our controller guide covers checking this, and it is the first thing to look at before buying faster drives to solve a speed problem.
RAID mode and HBA mode
This controller can work two ways, and the choice matters.
RAID mode is the default — the controller manages arrays, presents virtual disks to the operating system, and handles rebuilds itself.
HBA or non-RAID mode passes drives through individually so software handles redundancy. This is what software-defined storage and some file systems require, because they want to see the physical drives and their health data directly.
Two things worth knowing. Switching modes is destructive to existing array configuration, so decide before you build. And in RAID mode the controller filters some drive health information, which is one reason software storage prefers pass-through — our RAID versus HBA guide covers the distinction properly.
Moving arrays between servers
A situation that comes up regularly and works better than people expect.
The array configuration is written to the drives themselves, not only to the controller. That means a set of drives moved to another server with a compatible controller is generally recognised as a foreign configuration that can be imported.
Three cautions.
Move all drives from the array together, and keep their order recorded.
Controller firmware levels should be compatible. A large gap between source and destination can cause problems.
Have a backup before importing. This usually works and it is not something to attempt without a fallback — our backup guide covers why an untested backup is an assumption.
This is also why a failed controller is frequently recoverable: fit a compatible replacement, import the foreign configuration, and the data is intact.
Drive compatibility
Two points.
PERC controllers check drive firmware. A drive without Dell coding may work, may be flagged in iDRAC, may not report predictive failure warnings, or may be refused during a rebuild — which is the worst moment to find out. Our Dell drive guide covers this.
Bay capability is separate from the controller. What interfaces your bays support is a backplane decision, and the cable between controller and backplane is chassis-specific. Our backplanes guide and cabling guide cover both.
Choosing a RAID level with it
The controller supports several; the capacity of your drives should decide.
At modern drive capacities, a rebuild reads every remaining drive in full and takes many hours during which the array has no redundancy and every surviving drive is under sustained load — which is when second failures arrive.
So single parity becomes uncomfortable as capacity grows, and double parity is the sensible default on large drives. Our RAID guide covers the trade-offs, and the point worth repeating: RAID is not backup. It protects against a drive failing, not against deletion, corruption or the machine failing.
Before you order
Your PowerEdge model and generation. Whether you need the controller alone or also the cache battery module, which is a separate part with its own life. What cabling exists between controller and backplane. Whether you intend RAID mode or pass-through, since that affects what to buy. And if you are replacing a failed controller holding an existing array, tell us — firmware compatibility matters for importing the configuration.
Common questions
My server suddenly got slow with no errors. What should I check?
The controller cache battery. When it fails, the controller switches from write-back to write-through and write performance collapses — with no failed drive, no array error and nothing obviously broken. Check this before buying faster drives.
Why is my server slow only occasionally?
A controller relearning its cache battery temporarily drops to write-through, which produces periodic slowness on a schedule. That is normal behaviour rather than a fault, though a battery approaching end of life relearns more often.
Can I move an array to a different server?
Usually yes. The configuration is written to the drives, so a compatible controller recognises them as a foreign configuration that can be imported. Move all drives together, keep their order recorded, check firmware compatibility and have a backup first.
Should I run RAID mode or pass-through?
RAID mode for conventional arrays. Pass-through where software-defined storage or a file system wants to see physical drives and their health data directly, since RAID mode filters some of that. Decide before building — switching modes is destructive to existing configuration.
Is the cache battery included?
It is a separate part with its own limited life, so check whether it is included and what condition it is in. Treat it as a consumable on a replacement schedule rather than as a component that lasts the life of the controller.
Send us your PowerEdge model and whether you need the cache module, and tell us if you are replacing a controller holding an existing array.
