CMR (Conventional Magnetic Recording) and SMR (Shingled Magnetic Recording) are two different hard drive technologies that use different methods to store data on a disk platter. CMR is the traditional technology that has been used for decades. SMR is a newer technology developed in the last 10 years or so as a way to increase hard drive capacities. The key difference is how the tracks of data are arranged on the disk platters. In CMR drives, the tracks are discrete and separated, while in SMR drives, the tracks overlap like shingles on a roof. This allows SMR drives to squeeze more data onto each platter. However, it comes with some performance tradeoffs compared to CMR. In this article, we’ll take a closer look at how CMR and SMR drives differ and whether you can mix the two types in a system.
What is CMR?
CMR stands for conventional magnetic recording. It is the traditional and most common hard drive recording technology that has been used for decades. CMR drives write data sequentially across the magnetic tracks on the platters inside the hard drive.
With CMR, each track is separated by a physical boundary. This allows tracks to be individually written to without interfering with data on adjacent tracks. The physical separation prevents overlapping data, ensuring reliability and performance.
CMR drives have predictable performance because data is written sequentially. They also avoid problems like write errors and data corruption that can occur when tracks overlap on drives like SMR. Overall, CMR provides consistent performance and is considered the most reliable hard drive technology.
According to KitGuru’s review of the WD Red Pro 22TB HDD, CMR drives like this combine advanced features for optimal reliability and speed in NAS environments.
What is SMR?
SMR stands for shingled magnetic recording, a storage technology used in some hard disk drives (HDDs). In SMR drives, the tracks on the platters are partially overlapped or “shingled” to increase storage density. This allows SMR drives to offer more storage capacity than conventional drives using perpendicular magnetic recording (PMR) technology.
The key difference with SMR is that tracks are written in a sequential manner and new writes can potentially overwrite part of an adjacent track. To avoid this, the drive needs to first read and rewrite existing data to make room before new data can be appended sequentially. This can lead to performance issues compared to PMR drives, especially with random writes.[1]
Overall, SMR allows HDD vendors to increase capacities while reducing cost. But the shingled writing method also introduces complexities that require SMR drives to use special firmware to manage write operations. This can negatively impact performance in some workloads.
Comparing CMR and SMR
CMR (conventional magnetic recording) and SMR (shingled magnetic recording) drives differ in some key ways when it comes to performance, cost, and longevity:
Performance: CMR drives allow tracks of data to be written side-by-side without overlapping. This makes them better for random writes. SMR drives write new tracks that overlap part of the previously written track, resulting in slower write speeds, especially for random writes (Source).
Cost: SMR drives are generally less expensive than comparable CMR drives because their shingled writing allows more data to be stored per platter (Source).
Longevity: The overlapping tracks on SMR drives can lead to earlier failure compared to CMR. However, SMR drives are still generally reliable for normal usage. CMR may have an advantage for highly intensive workloads.
In summary, CMR drives are better for performance while SMR drives offer lower costs. Both can provide reliable longevity for typical usage.
Can You Mix CMR and SMR Drives?
Technically, it is possible to mix CMR and SMR drives in the same storage pool or RAID array. However, this is generally not recommended due to significant challenges it can cause for performance and data integrity.
One of the biggest issues with mixing drive types is that it can lead to uneven performance across the array. Writes to SMR drives are much slower than CMR, so write operations get bottlenecked waiting for the SMR drive to catch up. This leads to inconsistent latency and lower overall throughput for the entire array.
Another problem arises during rebuild operations. If a CMR drive fails and needs to be replaced, the RAID controller has to rebuild the data on the new drive. This involves heavy sustained writes that SMR drives struggle with. The rebuild process takes much longer and introduces risk of timeout failures. According to this source, it’s better to add an SMR drive to a degraded RAID than no drive at all, but performance will still suffer.
In addition, mixing drive types increases management overhead. With separate pools for each type, you can optimize the file system and caching for CMR vs SMR. When combined, finding the right balance is difficult.
For the best performance and data protection, it’s recommended to use either all CMR or all SMR drives in a single pool or array. The challenges and drawbacks typically outweigh the potential cost savings of mixing drive types.
Using CMR for OS/Apps
Dedicating CMR drives for your operating system and applications provides several key benefits:
Faster performance – CMR drives have better random read/write speeds which improves boot times and application launch speeds (History Computer). This makes the overall system feel much more responsive.
Reliability – The OS and apps benefit from the more reliable sequential writing of CMR. There is less risk of impacting performance through shingled writing like with SMR (MiniTool).
Compatibility – Most operating systems and applications are designed and optimized for CMR drives. Using CMR helps avoid any potential compatibility issues.
Longevity – The intense random writes of OS and apps is better suited for CMR drives over time. SMR drives used for this purpose may experience decreased lifespan.
Overall, dedicating fast CMR drives for your critical operating system and application files improves performance, reliability, compatibility, and longevity of your system.
Using SMR for Data
SMR drives can provide benefits for bulk data storage thanks to their higher storage densities and lower costs compared to CMR drives. The shingled write technology allows SMR drives to achieve much higher capacities in the same 3.5-inch hard drive form factors as CMR drives. For example, SMR drives are available at up to 20TB capacities while most CMR drives top out at 16TB.
This increased storage density comes with tradeoffs in performance, but for bulk secondary storage the high capacities and lower cost per terabyte can make SMR drives a good fit. Storing large archives, media libraries, backups, and other data that is infrequently accessed is a common use case. The downsides of SMR are less noticeable for this type of cold storage usage.
For high capacity bulk storage, SMR drives provide a cost-effective solution while leveraging their strenghts. Focusing SMR drives on data storage minimizes the impact of their shortcomings compared to CMR. As long as the tradeoffs are understood, SMR drives can be quite beneficial for maximizing storage density on a budget.
Maximizing Performance
When mixing CMR and SMR drives in a single array, it’s important to maximize performance by using a tiered storage approach. The key is to utilize the strengths of each drive type.
CMR drives offer better sustained write performance, making them ideal for storing frequently accessed data like operating systems, applications, databases, and active projects. Their consistent write speeds allow CMR drives to excel as boot drives or in caching roles.
Meanwhile, SMR drives provide high storage density at a lower cost. Though not optimized for sustained random writes, SMR drives work well for storing large amounts of “cold” data that is accessed infrequently. Media archives, backups, and other reference data are perfect examples.
Using caching and/or tiered storage software, you can place the most active data subsets on CMR drives. The SMR drives store the bulk of your data at a lower cost per TB. Overall performance remains strong since the most frequently accessed data resides on CMR drives.
RAID can also help maximize performance when combining CMR and SMR drives. RAID levels like RAID 10 provide redundancy while also improving write speeds for random I/O. Rebuilding RAID arrays with mixed drive types does take longer than uniform arrays however.
With proper planning, CMR and SMR drives can complement each other in a mixed array. But for optimal, consistent performance, matching drive types is still recommended.
Caveats and Considerations
When mixing CMR and SMR drives, there are some potential downsides and things to watch out for:
Performance can suffer significantly if too many SMR drives are used, as they have much slower write speeds than CMR drives. The more SMR drives in the array, the greater the performance hit (https://www.reddit.com/r/synology/comments/w6yz6e/smr_and_cmr_mix_in_shr2_array_major_issues_or/).
SMR drives can cause the array to rebuild much slower after a failure, prolonging the degraded state. The controller needs to deal with the shingled writing of the SMR drives which adds latency (https://community.synology.com/enu/forum/1/post/154929).
If the array is constantly being written to with large sequential writes, this can quickly fill up the SMR drive’s caching/buffer and choke performance. SMR drives are best suited for infrequent large writes.
There can be compatibility issues between RAID setups and the SMR drives, causing crashes or array failures. Always check compatibilty before mixing drive types.
In general, mixing drive types is not recommended if peak performance is needed. While feasible, there are scenarios where the SMR drives can bottleneck the entire array.
Conclusion
In summary, CMR and SMR drives utilize different technologies that provide varying performance characteristics. CMR drives are generally preferable for most applications due to their consistent performance and reliability. However, SMR drives can potentially be utilized for certain data storage use cases where maximum performance is not critical.
The general recommendation is to avoid mixing CMR and SMR drives within the same RAID array. While technically possible in some scenarios, doing so can introduce performance bottlenecks and complications. The ideal configuration is to use CMR drives for operating systems, applications, and active data. SMR drives can be relegated to archival data storage if needed to maximize storage capacity. However, SMR drives should be used with caution and understanding of their potential downsides.
When deciding whether to utilize SMR drives, be sure to weigh the benefits of increased storage capacity against potential performance impacts. Test proposed mixed CMR/SMR configurations thoroughly before deployment. Overall, maximize CMR drives where performance matters, and limit SMR drives to cold data storage where maximum throughput is not required.