Full Compass Systems’ Post

We may need to start keeping an eye on the length of our blog posts – but when you're covering building a high-performance, highly available #Lustre solution, there’s a lot to dig into. In our latest post, we walk through the process of setting up a dual-node system with shared NVMe drives, focused on maximizing both performance and availability. We cover everything from hardware choices to the configurations that let you make the most of Lustre with xiRAID Classic 4.1. Whether you're building a new Lustre installation or looking to expand an existing one, this article provides a detailed roadmap for creating a cutting-edge, fault-tolerant parallel file system suitable for demanding high-performance computing environments. If you’re interested in deep technical insights into storage solutions, this post might answer a lot of the questions you’ve been asking. 🔗 Check it out here: https://lnkd.in/eBcbd_ym

🛠️Optimizing TrueNAS SCALE for Better Memory Utilization🛠️ As a tech enthusiast, I always strive to make the most out of the resources at hand its now my Dell Technologies T320 which i recently I worked on optimizing how much memory ZFS can allocate for its ARC (Adaptive Replacement Cache) on my TrueNAS SCALE setup compared to my old platform. On the left, you can see the ARC cache consuming 110GB, and on the right, it’s been fine-tuned to utilize 144GB, by maximizing the use of free RAM, I’ve ensured that available memory isn’t going to waste—even if my primary use case is backups, where random new data often fills the cache. While this optimization isn’t critical for my setup, every little bit of efficiency counts in the long run. It's these tweaks that add up to a better-performing and more satisfying system. TrueNAS SCALE and ZFS enthusiasts have you tried similar optimizations? I’d love to hear about your experiences ⤵️ #truenas #zfs #optimization #learning #scale

Understanding Memory Card Labels: What You Need to Know Deciphering memory card labels can be challenging, but it's essential for choosing the right card for your needs. From form factors like SD and microSD to compatibility standards such as SDHC and SDXC, each label provides critical information. Learn more about how these labels can impact your device's performance and storage capacity. 🔗 Read the Full Article: https://zurl.co/mLnn #DataStorage #TechTips #MemoryCards

What are the challenges in bridging the hardware and software gap to optimize high-performance computing (HPC)? Join us for our broadcast, "HPC in Action," on Monday, June 24th, where we will address this question and more. Register today to get insights on practical examples of HPC deployment: https://okt.to/hpKP4E Jeff Wittich | Ampere #DCDbroadcast #HPC #ITWorkloads

We've recently updated our blog post on maximizing ZFS performance and evaluating xiRAID as a RAIDZ alternative. Our testing on 16 NVMe drives revealed that out-of-the-box ZFS performance starts at about 5.5 GB/s for both reads and writes. Through careful tuning of ZFS parameters and using a build with the Direct I/O patch, we boosted performance to 25.4 GB/s for writes and 30.6 GB/s for reads. The real breakthrough came when we replaced RAIDZ with xiRAID in a RAID6 configuration. This pushed write speeds to 47.4 GB/s and read speeds to 52.9 GB/s - nearly a 2x improvement. Notably, this also significantly reduced CPU utilization. Check out the full article for in-depth technical details and performance comparisons across different configurations: https://lnkd.in/gZeCZv7e #ZFS #StorageOptimization #HighPerformanceComputing #DataStorage

Data: PMax performance stable, but Search CPCs increasing https://lnkd.in/ea7qQwDi

📊 We're thrilled to announce the RESults from #HPC call 2024-2 📊 We have received a total of 243 petitions tu use our #RESources, from which a 48 of them have been ranked with an A due to their scientific and technical excellence. 🖥 Regarding CPU resources, we had available a total of 500M core hours, thanks to the large boost from our new machine #MareNostrum5. However, the total requested hours for this architecture was of 475M, not fully benefiting from this huge upgrade. We encourage our current and potential users to present more projects that are even more ambitious in means of CPU-based simulations in the next call. ⚡ In the other hand, regarding GPU accelerated #RESources, we offered a total of 58M hours, while the proposals reached an outstanding total of 179M hours, an oversubscription of 300%. We are really proud with this results and kindly encourage current and potential RES users to keep applying, even if the competition is high. 🇪🇺 Also remember to check EuroHPC Joint Undertaking (EuroHPC JU) calls, which also include GPU-accelerated resources from supercomputers from all around Europe: https://lnkd.in/dvmmXt_K

Comprehensive Guide to Banana Pi BPI-M7: Versatile Computing Power with Multiple Configurations https://lnkd.in/gCdHjx6E

New Year New storage solution for New NAS users! DRIVESTOR 2 Lite is equipped with a better CPU than the first generation. Also, it’s the first ARM NAS by us to support Btrfs, using DRIVESTOR 2 Lite to create Btrfs snapshots in Snapshot Center can make it easier to restore files after unintentional modifications. DRIVESTOR 2 Lite is an absolute bargain for NAS beginners https://lnkd.in/grrd-_Pr

As your backup needs grow, you can easily and affordably expand your RDX backup solution simply by adding more cartridges.RDX supports disk spanning so data can continue to grow across multiple cartridges. With a range of different capacities from 500 GB to 4 TB, there’s one that’s just right for you. And because RDX docks are forwards and backwards compatible, you won’t need to buy new hardware. https://lnkd.in/gQc48ayw