SPECstorage(TM) Solution 2020_ai_image Result WEKAIO, Inc. : WEKA - Public Cloud Reference SPECstorage Solution = 700 AI_Jobs (Overall Response Time = 0.85 msec) 2020_ai_image =============================================================================== Performance =========== Business Average Metric Latency AI_Jobs AI_Jobs (AI_Jobs) (msec) Ops/Sec MB/Sec ------------ ------------ ------------ ------------ 50 0.6 21752 4891 100 0.6 43505 9781 150 0.6 65258 14669 200 0.6 87011 19564 250 0.6 108764 24452 300 0.7 130517 29337 350 0.7 152269 34226 400 0.8 174022 39117 450 0.8 195775 44006 500 0.9 217528 48901 550 1.0 239281 53794 600 1.1 261034 58675 650 1.3 282787 63571 700 2.1 303042 68318 =============================================================================== Product and Test Information ============================ +---------------------------------------------------------------+ | WEKA - Public Cloud Reference | +---------------------------------------------------------------+ Tested by WEKAIO, Inc. Hardware Available January 2024 Software Available January 2024 Date Tested January 2024 License Number 4553 Licensee Locations Campbell, CA USA The WEKA; Data Platform is purpose-built to seamlessly and sustainably deliver speed, simplicity, and scale that meets the needs of modern enterprises and research organizations without compromise. Its advanced, software-defined architecture supports next-generation workloads in virtually any location with cloud simplicity and on-premises performance. Solution Under Test Bill of Materials ===================================== Item No Qty Type Vendor Model/Name Description ---- ---- ---------- ---------- ---------- ----------------------------------- 1 64 Azure Azure Standard WEKA storage back ends - Azure Virtual L8sv3 Standard L8s v3 Virtual Machine (8 Machine vCPUs, 64 GiB memory, 12.5 Gbps networking) 2 16 Azure Azure Standard Ubuntu clients with WEKA POSIX- Virtual D64sv3 compliant filesystem driver - Azure Machine Standard D64s v3 Virtual Machine (64 vCPUs, 256 GiB memory, 30 Gbps networking) Configuration Diagrams ====================== 1) storage2020-20240205-00066.config1.png (see SPECstorage Solution 2020 results webpage) Component Software ================== Item Name and No Component Type Version Description ---- ------------ ------------ ------------ ----------------------------------- 1 WekaFS File System Version The WEKA filesystem (WekaFS) 4.2.7.64 redefines storage solutions with its software-only approach and addresses common storage challenges by removing performance bottlenecks, making it suitable for environments requiring low latency, high performance, and cloud scalability. 2 Weka Backend Storage OS Ubuntu 20.04 Operating System (OS) for 50 Node storage virtual machines using Azure L8sv3 instances" 3 Clients Client OS Ubuntu 20.04 Operating System (OS) for 13 client virtual machines using Azure D64sv3 instances" Hardware Configuration and Tuning - Virtual =========================================== +----------------------------------------------------------------------+ | Component Name | +----------------------------------------------------------------------+ Parameter Name Value Description --------------- --------------- ---------------------------------------- Accelerated Enabled L8s instances have 12.5 Gbps and D64s Networking instances have 30 Gbps connectivity Hardware Configuration and Tuning Notes --------------------------------------- None Software Configuration and Tuning - Virtual =========================================== +----------------------------------------------------------------------+ | Networking | +----------------------------------------------------------------------+ Parameter Name Value Description --------------- --------------- ---------------------------------------- Jumbo Frames 3900 WekaFS recommended MTU for Azure. num_cores 3 WekaFS mount option to designate the number of frontend cores to allocate for the client. Software Configuration and Tuning Notes --------------------------------------- The single filesystem was attached via a single mount per client. The mount string used was "mount -t wekafs -o net=enP19658s5np0/10.0.0.56/20/10.0.0.1 -o net=enP28661s2np0/10.0.0.64/20/10.0.0.1 -o net=enP31350s3np0/10.0.0.63/20/10.0.0.1 -o num_cores=3 -o mgmt_ip=10.0.0.76 10.0.0.60/default /mnt/weka" The mount specifies that 3 WEKA front ends (cores) were used on the clients. Service SLA Notes ----------------- None Storage and Filesystems ======================= Item Stable No Description Data Protection Storage Qty ---- ------------------------------------- ------------------ -------- ----- 1 WEKA Filesystem with 88TB of usable WEKA 16+2 Local 64 storage distributed data 1.92TB protection NVMe Disk Number of Filesystems 1 Total Capacity 88 TB Filesystem Type WekaFS Filesystem Creation Notes ------------------------- Each storage backend has one Azure SSD attached to it, which are dedicated to the WEKA filesystem created . The storage cluster consisted of 64 storage servers with a single WEKA filesystem created in the cluster. Each client mounted the WEKA filesystem with a single mount. Storage and Filesystem Notes ---------------------------- None Transport Configuration - Virtual ================================= Item Number of No Transport Type Ports Used Notes ---- --------------- ---------- ----------------------------------------------- 1 12.5 Gbps 64 Used by WEKA Backend Server Nodes Virtual Nic 2 30 Gbps Virtual 16 Used by WEKA Frontend Client Nodes Nic Transport Configuration Notes ----------------------------- None Switches - Virtual ================== Total Used Item Port Port No Switch Name Switch Type Count Count Notes ---- -------------------- --------------- ------ ----- ------------------------ 1 Azure 12.5 Gbps 64 64 Used by WEKA Backend Ethernet with Server Nodes Accelerated Networking 2 Azure 30 Gbps 16 16 Used by WEKA Frontend Ethernet with Client Nodes Accelerated Networking Processing Elements - Virtual ============================= Item No Qty Type Location Description Processing Function ---- ---- -------- -------------- ------------------------- ------------------- 1 512 vCPU L8sv3 WEKA 2.8 GHz Intel(R) Xeon(R) WekaFS, Network Backend Platinum 8370C processors Communications, Storage Functions 2 1024 vCPU D64sv3 WEKA 2.8 GHz Intel(R) Xeon(R) WekaFS, Network Frontend Platinum 8370C processors Communications, Storage Functions Processing Element Notes ------------------------ None Memory - Virtual ================ Size in Number of Description GiB Instances Nonvolatile Total GiB ------------------------- ---------- ---------- ------------ ------------ Azure L8sv3 Virtual 64 64 V 4096 Machine Memory Azure D64sv3 Virtual 256 16 V 4096 Machine Memory Grand Total Memory Gibibytes 8192 Memory Notes ------------ None Stable Storage ============== WekaFS does not use any internal memory to temporarily cache write data to the underlying storage system. All writes are commited directly to the storage disk, and protected via WekaFS Distributed Data Protection (16+2 in this case). There is no need for any RAM battery protection. In an event of a power failure a write in transit would not be acknowledged. Solution Under Test Configuration Notes ======================================= The system under test consisted of 16 x D64sv3 Azure client virtual machines and 64 x L8sv3 backend virtual machines. Each client has a 30Gbps network connection and each WEKA backend has 12.5Gbps network connection. Other Solution Notes ==================== None Dataflow ======== None Other Notes =========== None Other Report Notes ================== The solution under test was a standard WekaFS cluster in dedicated server mode. The solution will handle both large file I/O as well as small file random I/O and metadata intensive applications. No specialized tuning is required for different or mixed use workloads. None of the components used to perform the test were patched with Spectre or Meltdown patches (CVE-2017-5754, CVE-2017-5753, CVE-2017-5715). =============================================================================== Generated on Thu Feb 15 10:31:28 2024 by SpecReport Copyright (C) 2016-2024 Standard Performance Evaluation Corporation