I have a RHEL7 box at work with a completely misconfigured partition scheme with XFS. Linux File System Comparison: XFS vs. I have 6 disks so I have created 3 logical disks, 2 SSDs each - just for testing. Here are some of those XFS RAID benchmarks up against Btrfs and. So it could be a. Btrfs El sistema de archivos Btrfs nació como. Ext4 is limited to a maximum file size of 16 TB, while NTFS can handle up to 256 TB worth of data. 79 1. This results in the clear conclusion that for this data zstd. Ext3 and Ext4 perform better on limited bandwidth (< 200MB/s) and up to ~1,000 IOPS capability. Although XFS is good, in practice I've found ext4 to be slightly faster. 21 merge window (now known as Linux 5. e. Looking at benchmarks however it seems to have poor. We benchmarked XFS vs EXT4 file system on these storage devices as well. The four hard drives used for testing were 6TB Seagate IronWolf NAS (ST6000VN0033-2EE) hard drives and the. So I installed a new Samsung 950 Pro NVMe SSD!! I previously had a Sandisk SSD formatted with ext4, just since it was the most stable (IMO) a few years back. The test results show that the Galaxy Note 10 performs better than the one plus 7 Pro in terms of random and SQLite write speed. However, Ext3 lacks advanced file system features. Note: Do not use mounted shared drives and any network file systems. ext4 also introduced delayed allocation of data, which adds a bit more risk with unplanned server outages while decreasing fragmentation and improving performance. I developed an application recently and compared the I/O performance of both and found ext4 to be slightly quicker for my application which was really just opening and reading whole files into memory. EXT4 is better in the general case. EXT4 vs. Stripe size and width. ZFS is an advanced filesystem and many of its features focus mainly on reliability. Conclusion. ZFS brings robustness and stability, while it avoids the corruption of large files. F2FS vs. 10 's new experimental ZFS desktop install option in opting for using ZFS On Linux in place of EXT4 as the root file-system, here are some quick benchmarks looking at the out-of-the-box performance of ZFS/ZoL vs. However, unlike Extended 4, it is not possible to disable journaling, thus it can be iffy to use on an SSD. 0 SSD for some reference data of the relative F2FS vs. This ext4 system has been in use for many years, so it is much improved from previous extensions and has greater bug removal support. 5 I/o scalability From day one, XFS has been designed to deal with high-performance disk subsystems, especially striped disk arrays with large aggregated bandwidth. Neither file system consistently outperforms the other in all workloads. SSD Filesystem: XFS vs F2FS vs Btrfs vs Bcachefs vs ext4 . It's an improved version of the older Ext3 file system. VM Memory and VCPU: Both VM’s have 2GB RAM and 1 VCPU of the same speed. XFS vs. 1829 tps). Basically, LVM with XFS and swap. F2FS vs. At the time, ZFS was significantly slower than xfs and ext4 except when the L2ARC was used. ext4 파일 시스템은 Red Hat Enterprise Linux 5에서 사용 가능한 기본 ext3 파일 시스템의 확장된 버전입니다. Running on an x570 server board with Ryzen 5900X + 128GB of ECC RAM. For those thinking of playing with Ubuntu 19. Then later, I was actually able to convert that from btrfs-raid10 to btrfs-raid1 overnight while in use. In. So logically, mainline Linux is more mature. As far as I know, the 4k block size is important for such webgui, it makes it faster to open sites (for ex. 5x faster than the common BSD UFS+J/UFS+S file-systems. 1 fell slightly short of the Linux file-system performance. EXT4 run a lot slower when we perform same SQL insert test; XFS respond a lot healthier at 2K INSERT + 2K UPDATE while EXT4 only have 59 for both. However, the performance of ZFS on FreeBSD/PC-BSD 8. Generally, ZFS is known for having great performance. 1-based Bcachefs-dev kernel. Momentum. Note that while these tests are not indicative of real-world performance, we can extrapolate these results and use this as one reason. I've done a good bit of Kernel dev for Android. 0 SSD for some reference data of the relative F2FS vs. ext4 is the safe choice that almost anyone. Snapshots, transparent compression and quite importantly blocklevel checksums. Migrating from ext4 to XFS" 3. User quotas for each shared folder. This is because BTRFS is optimized for handling small files, while EXT4 can struggle with multiple small files due to its delayed allocation. 7. – in the case of SATA/SSD, the ext4 scalability issue has an impact on tps rate after 256 threads and drop is 10-15%. 2. It can store large files and has advanced features as compared to Ext2 and Ext3. ZFS On Linux Benchmarks Storage : 2019-01-26: FreeBSD ZFS vs. After deciding to use LVM2 as volumemanager on our servers there was also the wish for an online resizeable filesystem. For more comprehensive coverage of performance improvements relating to storage and file systems, refer. It's not the most cutting-edge file system, but that's good: It means Ext4 is rock-solid and stable. Search Performance Test Btrfs Ext4 F2fs And Xfs On Linuxtrade goods, offerings, and more in your community area. • 2 yr. darkimmortal Member. NVMe drives formatted to 4096k. file-system comparison, here are some fresh benchmarks looking at the Btrfs, EXT4, F2FS, and XFS file-system benchmarks on a speedy WD_BLACK SN850 NVMe solid-state drive. Btrfs uses a checksum to ensure that the data doesn’t corrupt, on the other hand, Ext4 doesn’t ensure data integrity. Across the three tested RAID modes, EXT4 was performing the worst. - Linux Kernel 5. Various internet sources suggest that XFS is faster and better, but taking into account that they also suggest that EXT4 is. 7. The charts show sequential reads (top) and writes (bottom) on XFS (left) and EXT4. It would be interesting to see a new benchmark result of CoW filesystems BTRFS vs ZFS in real world 2022. Swap space. F2FS, XFS, ext4, zfs, btrfs, ntfs, etc. Because, firstly, it does not do data journalling or "ordered writing" and in a crash/reset you end up with random data (probably top secret files erased earlier) in your new files. The compression ratio of gzip and zstd is a bit higher while the write speed of lz4 and zstd is a bit higher. The NTFS support was powered by FUSE. 2020. Btrfs Benchmarks comparison, here is a wider look at mainline file-systems on the Linux 4. When running one copy of the SQLite embedded database library, the XFS file-system had a slim lead over NILFS2 and F2FS while EXT4 was the slowest on this Linux 5. the COW which saves alot of space and increases the speed. Built By the Slant team. the fact that maximum cluster size of exFAT is 32MB while extends in ext4 can be as long as 128MB. Vide. 5k tps, so ~20% increase), but the jitter is clearly much higher. At 32 threads ext4 was 28% faster (2345 tps vs. Compared to Ext4, XFS has a relatively poor performance for single threaded, metadata-intensive workloads. However, to fully exploit ext4's performance capabilities, files need to be restructured to use the extents storage mechanism, which isn't done automatically during the conversion. The observation was that XFS is useful when your machine has multiple cores and fast disk that XFS can utilize. XFS can sometimes detect the geometry under software RAID, but in case you reshape it or you. 파일 시스템. At 16 threads it was a draw (2036 tps vs. The host is proxmox 7. XFS provides a more efficient data organization system with higher performance capabilities but less reliability than ZFS, which offers improved accessibility as well as greater levels of data integrity. EXT4 vs. Various benchmarks have concluded that the actual ext4 file system can perform a variety of read-write operations faster than an NTFS partition. ) – improvements, bugfixes. Ext3 and Ext4 perform better on limited bandwidth (< 200MB/s) and up to ~1,000 IOPS capability. Differences Between Ext3/4 and XFS 4. 03. We looked into the performance of popular filesystems with this configuration. À partir de Red Hat Enterprise Linux 7. At 16 threads it was a draw (2036 tps vs. See Core dump#Disabling automatic core dumps. You can sometimes run into bugs and issues if your home directory is partitioned in XFS, BTRFS, or ZFS. First, btrfs is a perfectly cromulent single-disk ext4 replacement. 1. AIM7 Benchmark For those thinking of playing with Ubuntu 19. A backup strategy without data integrity protection from the file system or some other mechanism will blindly backup corrupted data if data corruption occurs. There was a higher risk than upon disconnection or loss of power than some of the files are truncated. I've built many (and maintain a number of) ZFS hosts with very large filesystems / databases. you can chroot, but you won't really have a performance issue with the native WSL drive. 2. At 64 threads ext4 was even 47% faster (2362 tps vs. EXT4 vs. , Ext4 or XFS): they present whole families of file systems. 1. 5. It's a 64-bit, journaling filesystem that has been built into the Linux kernel since 2001 and offers high performance for large filesystems and high degrees of concurrency (i. So I did two rounds: the. 4935 2026 MB/s. XFS File System. It is a rock-solid option since it has been around for long, bringing with it all the years of. 1. See Sysctl#Virtual memory for details. @Falzo said: I think in general the comparison is a bit. Optane SSD RAID Performance With ZFS On Linux, EXT4, XFS, Btrfs, F2FS Storage : 2019-06-20: Linux 5. All these benchmarks were carried out in a fully-automated and. SGI created XFS to handle huge files (xxx MB or more) very well. 4. XFS and EXT4 are common low-overhead / performance options, btrfs. ago. XFS is optimized for large file transfers and parallel I/O operations, while ext4 is optimized for general-purpose use with a focus on security. 9, 97. Generally, ZFS is known for its superior performance in large-scale storage environments, while Btrfs is more performant in smaller-scale deployments. xfs man page for additional information) 1: Example /proc/mdstat file with missing device:XFS, like Ext4, is a journaling filesystem. With not having the time to conduct the usual kernel version vs. Another test: everything is the same, upgraded kernel to 5. Updating 1 million files takes ages. Ext4 파일 시스템. Off a Linux 5. Watching LearnLinuxTV's Proxmox course, he mentions that ZFS offers more features and better performance as the host OS filesystem, but also uses a lot of RAM. F2FS vs. all kinds for nice features (like extents, subsecond timestamps) which ext3 does not have. It provides good performance with SSD and supports the TRIM (and FITRIM) feature to keep good SSD performance over time (this clears unused memory blocks for quick later write access). This is addressed in this knowledge base article; the main consideration for you will be the support levels available: Ext4 is supported up to 50TB, XFS up to 500TB. IMO XFS and F2FS seem like good choices for the most performance (F2FS was designed for SSDs). 86 1. Given Canonical has brought. 但无论如何，各个文件系统都需要存储这三类信息，因为这是内核规定的（见下）。. #filesystem #ext4 #xfs #linuxExplicación de las diferencias entre sistemas de archivos, en este vídeo se comparan los 2 mas usados en GNU/Linux. ext4 and also reiserfs store files in a different way. if date corruption from power loss is an issue with btrfs. Main features: Data protection features, including snapshot, replication, and point-in-time recovery. xfs man page for additional information) 1: Example /proc/mdstat file with missing device: It uses mount point into /var/lib/longhorn with a standard filesystem (ext4 or xfs). Yes, both BTRFS and ZFS have advanced features that are missing in EXT4. Ticket Spinlocks. For anything with higher capability, XFS tends to be faster. 64-Bit Support 2. But I was more talking to the XFS vs EXT4 comparison. It appears that ZFS may be a viable option, but do bear in mind to disable compression and encryption as they may impact performance. EXT4: 2. 4% utilization. ext4 to specify a file system label. e. 0 solid state drives using other file-systems -- including EXT4, XFS, and Btrfs with Linux 3. 6. CoW filesystems like BtrFS are great and full of advantages, but the performance drop away from XFS is notable. 8 testing. The per-second throughput varies roughly between 5k and 9k tps—not great, not terrible. In Summary, ZFS, by contrast with EXT4, offers nearly unlimited capacity for data and metadata storage. It was mature and robust. For large sequential reads and writes XFS is a little bit better. Edit: fsdump / fsrestore means the corresponding system backup and restore to for that file system. 5. Red Hat Enterprise Linux 6 Performance Features" 2. For storage, XFS is great and sometimes has higher performance than EXT4. Use the -L flag of mkfs. We may have lengthy talk on ext vs XFS vs f2fs and btrfs vs zfs and there are many more points to be mentioned, but for regular users. Abstract and Figures. XFS Storage : 2019-01-07: Linux RAID Benchmarks With EXT4 + XFS Across Four Samsung NVMe SSDs Storage : 2018-08-24: Reiser4 File-System Benchmarks With Linux 4. So its ext4. The Ext4 File System. 34, NO. It has lower performance than tried and true ext4 but that is the cost to pay for the features it has. For this reason, I took the time to extend the same benchmark to Oracle ASM (Automatic Storage Management) and also to Oracle Enterprise Linux (OEL). ext4: 1 1 Toshiba. Continue readingWindows has always been terribly slow to update, say, all file permissions in a large directory structure. And then I have formatted them with ext4, XFS and BTRFS. For single disks over 4T, I would consider xfs over zfs or ext4. I chose two established journaling filesystems EXT4 and XFS two modern Copy on write systems that also feature inline compression ZFS and BTRFS and as a relative benchmark for the achievable compression SquashFS with LZMA. XFS (2002) – originally SGI Irix 5. XFS, EXT4, and BTRFS are file systems commonly used in Linux-based operating systems. Over time, these two filesystems have grown to serve very similar needs. 3. These days, you just pick the filesystem you need for the device. File-systems tested on the NVMe SSD included Btrfs, EXT4, F2FS, XFS, and NTFS. Improve this answer. The only case where XFS is slower is when creating/deleting a lot of small files. but rather comparable to the usage of md-raid underneath or LVM. But if you're hoping to replace ZFS—or a more complex stack built on discrete RAID management, volume management, and simple. I ran performance benchmarks comparing XFS with EXT4 for MongoDB on AWS EC2 to find out exactly what you were wondering about. 1. However, Linux limits ZFS file system capacity to 16 tebibytes. ago. XFS does not require extensive reading. XFS is a high-performance, journaling file system designed for high scalability. XFS vs EXT4. For example, an XFS file system's size can be increased, but it cannot reduced. Agree, actually I have a bunch of freebsd for ZFS. I used to format XFS using mkfs. With Btrfs you get self healing, snapshots, copy on write, background file system checks, online defragmentation, and much more. ZFS's biggest disadvantage in my opinion is memory usage: If you have less than 16 GiB of RAM for a production server, you may want to. NTFS Benchmarks Continuing on from yesterday's Linux 4. 14 file-system performance comparison with a traditional hard drive. BTRFS is newer, and the performance is not as good in many cases, but it is not far off. XFS is better in general with WT, as the MongoDB production notes suggest. XFS scales much better on modern multi-threaded workloads. When I use ext4 the 4k speed is 5-7 MB/s. Sure the snapshot creation and rollback ist faster with btrfs but with ext4 on lvm you have a faster filesystem. Recent improvements to the XFS file system have shown it to have the better performance characteristics for Kafka’s workload without any compromise in stability. I tested an XFS filesystem on an LVM physical volume vs. The fuse and fuseblk file system types are different from traditional file systems (e. Common Commands for ext3 and ext4 Compared to XFS. For personal and SOHO use, EXT4 is the most commonly used file system in Linux systems. With not having the time to conduct the usual kernel version vs. We use this almost exclusively where performance matters as the primary concern. Btrfs native RAID was much faster for sequential writes than EXT4/XFS on Linux Software RAID. It scales with a number of controller replicas, which can bring extra. Compared to XFS, Ext4 handles less file sizes for example maximum supported size for Ext4 in RHEL 7 is 16TB compared to 500TB in XFS. Phoronix: Linux 5. Replica set members can definitely use different filesystems -- members aren't even aware of what filesystems are in use by their peers. It has been suggested that ZFS may not be optimal for fread/fwrite operations, and it may be advisable to utilize ZFS for non-root directories while utilizing ext4 for the remainder of the system for optimal. ext4 can claim historical stability, while the consumer advantage of btrfs is snapshots (the ease of subvolumes is nice too, rather than having to partition). Unfortunately Synology uses ext4 and btrfs; no support for xfs out of the box. I will use Ext4 until something more viable with at least the same level of stability takes its place. To achieve expected performance by tweaking the IRQ affinity, consider few important parameters like Linux handling of the server topology, NIC driver stack, default. ), the better for efficient disk usage, in case there's a lot of small files on that partition. 1. – in the case of SATA/SSD, the ext4 scalability issue has an impact on tps rate after 256 threads and drop is 10-15%. e. XFS was originally developed by Silicon Graphics for IRIX and later ported to Linux. Ext4#Improving performance and XFS#Performance. Honestly I wasn't aware of the huge amount of extends still created - that explains a bit. Ext3 and Ext4 perform better on limited bandwidth (< 200MB/s) and up to ~1,000 IOPS capability. Also BRTFS compresses the file system using less space compared to EXT4 but again the tradeoff is it uses more computer. 21 merge window (now known as Linux 5. 1. EXT4 lacks more robust features but is stable and well-supported on all Linux operating systems. This is the first time that the new EXT4 and Btrfs and NILFS2 filesystems have been directly compared when it comes to their disk performance though the results may surprise. Besides the XFS/EXT4/F2FS tests on the Western Digital hard drive, I also repeated the tests on a Samsung 860 QVO 1TB SATA 3. I used a Dell R630 machine with two E5-2699 CPUs in it. 0 File-System Benchmarks: Btrfs vs. 1, 4. Use the storage driver with the best overall. When running MongoDB in production on Linux, you should use Linux kernel version 2. 4 To 4. If we apply a fix by mounting ext4 with dioread_nolock or use xfs, throughput looks good. 3. 1. Adding an LVM layer actually reduces performance a tiny bit. Back when Bcachefs debuted in. 3 with zfs-2. XFS Written by Michael Larabel in Storage on 7 January 2019. Multimedia Sanctuaries: With large files as daily bread, ext4 is indispensable. Also, server raid originally md raid5 (4x4TB NAS drives) with XFS had taken all day to build, but creating btrfs-raid10 was seconds. On SSDs and HDDs, it delivers fast atomic actions and stable values in the IOzone benchmark. Figure 3 - Using psync engine with FIO* tool. With Bcachefs on its trek towards the mainline Linux kernel, this week I conducted some benchmarks using the very latest Bcachefs file-system code and compared its performance to the mainline Btrfs, EXT4, F2FS, and XFS file-system competitors on both rotating and solid-state storage. 0-050600-generic. With a decent CPU transparent compression can even improve the performance. Increased Performance of ext4 vs. g. That's disgusting enough for me not to want it. To. exFAT vs NTFS. BTRFS also had somewhat higher latency than EXT4, meaning that it took longer for files to be accessed on the file system. Each of the five file-systems were tested on the same NVM Express SSD from the Linux 4. 8. file-system comparison, here are some fresh benchmarks looking at the Btrfs, EXT4, F2FS, and. It presents the. They’re fast and reliable journaled filesystems. • A specification for accessing solid-state drives (SSDs) attached through the PCI Express (PCIe) bus. Migrating from ext4 to XFS" Collapse section "3. resource utilization; finally, the impact of. BTRFS. But not enough users follow the guide on and instead do stuff that actually makes the system worse. With 4K random reads by FIO, the SATA/USB performance was flat across. Thus, if those who rely on CPU-bound workload with little concurrency work better and faster using Ext3 or Ext4. It is suitable for PC platforms and. Benchmark of Ext4, XFS, Btrfs, ZFS With PostgreSQL Database benchmark on a VPS, using several filesystem and configuration options. Each volume is like a single disk file. As you can imagine there is not a single and. 0 while today is just a comparison of six file-systems using a traditional HDD. Btrfs is one of the most popular newly created file systems, and was. 6. ext4 is the successor to ext3. 4 usage of the XFS file system. It was first released in 2008 and serves as the successor to ext3. • PCIe SSD devices designed based on the NVMe specification are called NVMe-based PCIe SSD’s • Provides a scalable host controller interface for devices in various form. EXT4 I have no experience with, but XFS, despite all the hype, I think is better avoided. Both cases, a mechanical drive. xfs -l size=64m (notes fromHas anyone compared the IO performance of WSL2 "emulated disk" vs a bare physical disk from wsl --mount --bare ? (Is there any comparison of ext4 vs XFS vs ZFS? I will run fio myself but I'd like to compare benchmarks to avoid wasting too much time). 15 FS performance to Linux 3. The file-systems being benchmarked here are EXT4, XFS, and Btrfs. Short answer: under GNU/Linux, you should use a GNU/Linux native file system, such as ext4, XFS or btrfs, as your root partition, for stability and security. 7. Sequential reads, however, were coming in slower. Here are some more benchmarks. Page 1 of 4. It has been suggested that ZFS may not be optimal for fread/fwrite operations, and it may be advisable to utilize ZFS for non-root directories while utilizing ext4 for the remainder of the system for optimal. Increased Performance of ext4 vs. Though not as large of a difference when comparing to an SD card. Features of the XFS and ZFS. XFS, EXT4) have better tools available for Linux, for recovery and maintenance, and probably a more complete implementation. Prior to EXT4, in many distributions, EXT3 was the default file-system. DragonFlyBSD HAMMER2 vs. To me this looks like the best option in terms of performance, though it doesn't appear to be a popular choice -- reading the documentation, as well as discussions in various threads here I only see most users debating about NFS vs SMB vs iSCSI. Some file system repairs have demonstrated up to a six-fold increase in performance. Optane SSD RAID Performance With ZFS On Linux, EXT4, XFS, Btrfs, F2FS Storage : 2019-06-20: FreeBSD ZFS vs. Filesystem benchmarks with EXT4, XFS and ZFS | GCore GmbH Linux filesystem benchmarks EXT4, XFS and ZFS compared START Help Filesystems Home. An anonymous reader writes "Phoronix has published Linux filesystem benchmarks comparing XFS, EXT3, EXT4, Btrfs and NILFS2 filesystems. EXT4 has entirely different design goals, none of which are data integrity. ZoL Performance, Ubuntu ZFS On Linux Reference Storage : 2019-04-24: FreeBSD ZFS vs. In the case of the Intel 900p SSD, the XFS results were too fast to accurately measure while EXT4 and F2FS took just two seconds to complete while Btrfs took six seconds. > Last time I ran these tests, xfs and ext4 pulled very similar results, > and both were miles ahead of btrfs. Linux EXT4/Btrfs RAID With Twenty SSDs. 0-050600-generic. Up to 8 threads xfs was few percent faster (~10% on average). 3. The Infortrend RAID is a 24-disk box arranged as two RAID-6 arrays of 12 disks each, each disk 1 TB. 15 kernel was unchanged compared to Linux 3. brown2green. This of course comes at the cost of not having many important features that ZFS provides. 38 We see that on the SMR disk btrfs has most of the advantage on overall ops that it has on ext4, but. I'm pretty sure some of the higher performance ones. English Table of Contents Types of File Systems Local File Systems Overview The XFS File System The Ext File System Family Ext4 File System Choosing a Local File System. Its also not aligned with the Stratis concept, as that is closer to thin LVM with XFS just providing the top layer. XFS uses one allocation group per file system with striping. Performance is a QCOW2 vs RAW thing, not ext4 vs LVM (which adds another layer on top of ext4). As Microsoft makes more progress with ReFS on Windows 11, Linux is also getting performance optimizations and improvements on some of its major file systems, namely, F2FS, Btrfs, and EXT4. Btrfs vs. A few days ago I ran some fresh hard drive file-system benchmarks on Linux 4. ZFS 101—Understanding ZFS storage and performance. ext4 can claim historical stability, while the consumer advantage of btrfs is snapshots (the ease of subvolumes is nice too, rather than having to partition). 0 mainline kernel and using the stock mount options. 7. ZFS has built-in RAID support with various RAID-Z levels (RAID-Z, RAID-Z2, and RAID-Z3). Small example: One plus 7 Pro has the same UFS 3. So for a large video collection, I think I will stick with ext4 still. When XFS was designed, “high performance” meant a. A number of Phoronix readers have been asking about some fresh file-system comparisons on recent kernels. g. If you need to use it cross-platform you should probably go with either NTFS or ExFAT. If this filesystem will be on a striped RAID you can gain significant speed improvements by specifying the stripe size to the mkfs. First of all, some background history. Perhaps btrfs is much better for SSDs, but in. With not having the time to conduct the usual kernel version vs. The XFS one on the other hand take around 11-13 hours!ZFS vs EXT4 for Host OS, and other HDD decisions. See below: XFSYou're welcome. Although use of the Ext4 filesystem is one possibility for performance issues with MongoDB and WiredTiger (particularly under significant write load), there may be other issues affecting your use case. In terms of XFS vs Ext4, XFS is superior to Ext4 in the following aspects: Larger Partition Size and File Size: Ext4 supports partition size up to 1 EiB and file. 7 - EXT4 vs. 0 and today those results are being complemented by the solid-state drive results. 61 CommentsIn some ways, btrfs simply seeks to supplant ext4, the default filesystem for most Linux distributions. In our experience Kafka is known to have index failures on such file systems. F2FS vs. The last time I benchmarked them they were very close, with some differences for specific circumstances: XFS open() and readdir() remained fast as the number of files in a directory grew very large (tens of thousands or hundreds of thousands) whereas EXT4 performance degraded. I've read and have anecdotally (not scientific and could be affected by other things) experienced Btrfs being slower than ext4. Also, I found out the sysbench benchmark I used at the time was not a fair choice since the dataset it generates compresses much less than a. These days, you just pick the filesystem you need for the device.