SPECsfs2008_nfs.v3 Result ================================================================================ Avere Systems, Inc. : FXT 3800 (3 Node Cluster, ZFS Storage Config) SPECsfs2008_nfs.v3 = 180043 Ops/Sec (Overall Response Time = 0.87 msec) ================================================================================ Performance =========== Throughput Response (ops/sec) (msec) --------------------- ---------------- 17808 0.3 35658 0.4 53519 0.5 71383 0.6 89274 0.7 107189 0.8 125289 1.0 143053 1.2 160709 1.7 180043 2.7 ================================================================================ Product and Test Information ============================ Tested By Avere Systems, Inc. Product Name FXT 3800 (3 Node Cluster, ZFS Storage Config) Hardware Available July 2013 Software Available March 2014 Date Tested March 2014 SFS License Number 9020 Licensee Locations Pittsburgh, PA USA The Avere Systems FXT 3800 Edge filer running AvereOS V3.2 provides NAS storage that enables performance scalability at the edge while leveraging both Legacy NAS and object-based Cloud storage capacity at the core. The AvereOS Hybrid NAS software dynamically organizes hot data into RAM, SSD and SAS tiers, retaining active data on the FXT Edge filer, and placing inactive data on the Core filer. The FXT Edge filer managed by AvereOS software provides a global namespace, clusters to scale to as many as 50 nodes, supports millions of I/O operations per second, and delivers over 100 GB/s of I/O bandwidth. The FXT 3800 is built on a 64-bit architecture that provides sub-millisecond responses to NFSv3 and CIFS client requests consisting of read, write, and directory/metadata update operations. The AvereOS Local Directories functionality enables the Avere Edge filer to immediately acknowledge all filesystem requests, and flush inactive directory data and user data to the Core filer. The tested Edge filer configuration consisted of (3) FXT 3800 nodes. The system also included (1) OpenIndiana ZFS server in the role of a single Core filer mass storage system exporting a single filesystem. A WAN simulation kernel module was configured into AvereOS to introduce 150 milliseconds of WAN latency (75 milliseconds in each direction) for all communications with the Core filer, thus simulating a cloud storage infrastructure. In this configuration, Avere's AOS presents a single filesystem to all clients and provides instantaneous responses to client operations, despite the 150 millisecond network roundtrip times present with all communications to the Core filer. Configuration Bill of Materials =============================== Ite m Vendo Model/ No Qty Type r Name Description --- --- ---- ----- ------- ----------- 1 3 Storage Avere FXT Avere Systems Hybrid NAS Edge filer appliance Applian Syste 3800 running Avere OS V3.2 software. Includes (13) 600 ce ms, GB SAS Disks and (2) 400GB SSD Drives. Inc. 2 1 Server Genst CSRM- 4U 24-Drive Chassis for OpenIndiana Core filer Chassis or 4USM24A storage NFS Server. SRA 3 1 Server Genst MOBO- Supermicro X8DTE Motherboard for OpenIndiana Core Motherb or SX8DTEF filer storage NFS Server. oard 4 2 CPU Genst CPUI- Intel Westmere Quad Core 2.4 GHz 12MB L3 CPU for or X5620E OpenIndiana Core filer storage NFS Server. 5 2 CPU Genst CPUI- SM 2U 3U P0038P Heat Sink for OpenIndiana Core Heat or 2USMP00 filer storage NFS Server. Sink 38P 6 8 Memory Arrow MT36JSZ Micron DRAM Module DDR3 SDRAM 8GByte 240 RDIMM for F1G72PZ OpenIndiana Core filer storage NFS Server. -1GD1 7 1 Disk Seaga ST500NM Seagate 500GB Constellation ES SATA2 System Disk Drive te 0011 for OpenIndiana Core filer storage NFS Server. 8 9 Disk Seaga ST4000N Seagate 4TB Constellation ES.3 SATA2 Disks for Drive te M0033 OpenIndiana Core filer storage NFS Server. 9 1 RAID Co LSI LSI0020 LSI 4 internal/4 external port 6Gb/s SATA+SAS RAID ntrolle 9 card, x8 PCIe 2.0 host interface for OpenIndiana r Core filer storage NFS Server. 10 1 Network Intel E10G42B Intel 10 Gigabit AF DA Dual Port Server Adapter Card TDA for OpenIndiana Core filer storage NFS Server. Server Software =============== OS Name and Avere OS V3.2 Version Other Software Core filer storage server runs OpenIndiana SunOS 5.11 oi_151a, this package is available for download at http:// dlc.openindiana.org/isos/151a/oi-dev-151a-x86.iso Filesystem Avere OS V3.2 Software Server Tuning ============= Name Value Description ---- ----- ----------- Writeback Time 12 Files may be modified up to 12 hours before hours being written back to the Core filer storage system. buf.autoTune 0 Statically size FXT memory caches. buf.InitialBalance. 57 Tune smallFile buffer to use 57 percent of smallFilePercent memory pages. buf.InitialBalance. 10 Tune largeFile buffers to use 10 percent of largeFilePercent memory pages. cfs.randomWindowSize 32 Increase the size of random IOs from disk. cluster.dirMgrConnMult 12 Multiplex directory manager network connections. dirmgrSettings. 400000 Increase the directory log size. unflushedFDLRThresholdToStart 00 Flushing dirmgrSettings. 120000 Increase the directory log size. maxNumFdlrsPerNode 000 dirmgrSettings. 150000 Increase the directory log size. maxNumFdlrsPerLog 0 dirmgrSettings. 1 Balance directory manager objects across balanceAlgorithm the cluster. tokenmgrs.geoXYZ. no Disable the use of full control read fcrTokenSupported tokens. tokenmgrs.geoXYZ. no Disable token contention detection. trackContentionOn tokenmgrs.geoXYZ. 154000 Set a threshold for token recycling. lruTokenThreshold 00 tokenmgrs.geoXYZ. 155000 Set maximum token count. maxTokenThreshold 00 vcm.readdir_readahead_mask 0x3000 Optimize readdir performance. vcm. 1 Disable optimistic filehandle recycling. disableAgressiveFhpRecycle vcm.readdirInvokesReaddirplus 0 Disable optimistic trigger of client readdir calls to readdirplus fetches. initcfg:cfs.num_inodes 280000 Increase in-memory inode structures. 00 initcfg:vcm.fh_cache_entries 160000 Increase in-memory filehandle cache 00 structures. initcfg:vcm. 180000 Increase in-memory name cache structures. name_cache_entries 00 Server Tuning Notes ------------------- None Disks and Filesystems ===================== Numb er of D Usable Description isks Size ----------- ---- -------- Each FXT 3800 node contains (13) 600 GB 10K RPM SAS disks. All FXT 39 21.3 TB data resides on these disks. Each FXT 3800 node contains (2) 400 GB eMLC SSD disks. All FXT 6 2.2 TB data resides on these disks. Each FXT 3800 node contains (1) 250 GB SATA disk. System disk. 3 698.9 GB The Core filer storage system contains (9) 4 TB SATA disk. The ZFS 9 25.4 TB file system is used to manage these disks and the FXT nodes access them via NFSv3. The Core filer storage system contains (1) 500GB system disk. 1 465.7 GB Total 58 50.0 TB Number of 1 Filesystems Total Exported 22116 GiB (OpenIndiana Core filer storage system capacity) Capacity Filesystem Type TFS (Tiered File System) Filesystem Default on FXT nodes. Creation Options OpenIndiana Core filer storage server ZFS filesystem created with commands:   zpool create vol0 raidz2   zfs create -o quota=21.5t -o sharenfs=root=@0.0.0.0/0 vol0/cloudboi Filesystem Config 7D+2P RAID6 (RAIDZ2) configuration on OpenIndiana Core filer storage server, with 1 RAID set in the zpool. Fileset Size 20862.2 GB Network Configuration ===================== Number Ite of m Ports No Network Type Used Notes --- ------------ ------- ----- 1 10 Gigabit 3 One 10 Gigabit Ethernet port used for each FXT 3800 Ethernet Edge filer appliance. 2 10 Gigabit 1 The Core filer storage system is connected via 10 Ethernet Gigabit Ethernet with simulated latency of 150ms (75ms each direction.) Network Configuration Notes --------------------------- Each FXT 3800 was attached via a single 10 GbE port to one Gnodal GS7200 72 port 10 GbE switch. The load generating client was attached to the same switch. The Core filer storage server was also attached to the same switch. A WAN simulation software module on the FXT 3800 system was configured to inject 150ms (75ms in each direction) of WAN latency between the Avere Edge filer cluster and the Core filer. A 1500 byte MTU was used throughout the network. Benchmark Network ================= An MTU size of 1500 was set for all connections to the switch. The load generator was connected to the network via a single 10 GbE port. The SUT was configured with 3 separate IP addresses on one subnet. Each cluster node was connected via a 10 GbE NIC and was sponsoring 1 IP address. Processing Elements =================== Ite m No Qty Type Description Processing Function --- --- ---- ----------- ------------------- 1 6 CPU Intel Xeon CPU E5645 2.40 GHz FXT 3800 Avere OS, Network, NFS/ Hex-Core Processor CIFS, Filesystem, Device Drivers 2 2 CPU Intel Xeon E5620 2.40 GHz OpenIndiana Core filer storage Quad-Core Processor systems Processing Element Notes ------------------------ Each file server has two physical processors. Memory ====== Number of Description Size in GB Instances Total GB Nonvolatile ----------- ---------- ------------- -------- ----------- FXT 3800 System Memory 144 3 432 V Core filer storage system memory 64 1 64 V FXT 3800 NVRAM 2 3 6 NV Grand Total Memory Gigabytes 502 Memory Notes ------------ Each FXT node has main memory that is used for the operating system and for caching filesystem data. Each FXT contains two (2) super-capacitor-backed NVRAM modules used to provide stable storage for writes that have not yet been written to disk. Stable Storage ============== The Avere filesystem logs writes and metadata updates to the PCIe NVRAM modules. Filesystem modifying NFS operations are not acknowledged until the data has been safely stored in NVRAM. In the event of primary power loss to the system under test (SUT), the charged super-capacitor provides power to the NVRAM circuitry to commit the NVRAM log data to persistent flash memory on the NVRAM card. The NVRAM log data is indefinitely persisted to on-board flash memory. When power is restored to the SUT, the NVRAM log data is restored from the persistent on-board flash to the NVRAM subsystem, and during startup, AvereOS replays this log data to the on-disk filesystem thus assuring a stable storage implementation. The above described implementation handles cascading power failures in the same manner: persist NVRAM to stable flash upon power failure, restore NVRAM log data from flash and replay logs to filesystem upon power-on. System Under Test Configuration Notes ===================================== The system under test consisted of (3) Avere FXT 3800 nodes. Each node was attached to the network via 10 Gigabit Ethernet. Each FXT 3800 node contains (13) 600 GB SAS disks and (2) 400GB eMLC SSD drives. The OpenIndiana Core filer storage system was attached to the network via a single 10 Gigabit Ethernet link. The Core filer storage server was a 4U Supermicro server configured with a software RAIDZ 7+2 RAID6 array consisting of (9) 4TB SATA disks. Other System Notes ================== N/A Test Environment Bill of Materials ================================== Item No Qty Vendor Model/Name Description ---- --- ------ ---------- ----------- 1 1 Supermic SYS-1026T- Supermicro Server with 48GB of RAM running CentOS ro 6RFT+ 6.4 (Linux 2.6.32-358.0.1.el6.x86_64) 2 1 Gnodal GS7200 Gnodal 72 Port 10 GbE Switch. 72 SFP/SFP+ ports Load Generators =============== LG Type Name LG1 BOM Item # 1 Processor Name Intel Xeon E5645 2.40GHz Hex-Core Processor Processor Speed 2.40 GHz Number of Processors (chips) 2 Number of Cores/Chip 6 Memory Size 48 GB Operating System CentOS 6.4 (Linux 2.6.32-358.0.1.el6.x86_64) Network Type Intel Corporation 82599EB 10-Gigabit SFI/SFP+ Load Generator (LG) Configuration ================================= Benchmark Parameters -------------------- Network Attached Storage Type NFS V3 Number of Load Generators 1 Number of Processes per LG 768 Biod Max Read Setting 2 Biod Max Write Setting 2 Block Size 0 Testbed Configuration --------------------- Networ Target LG No LG Type k Filesystems Notes ----- ------- ------ --------------- ----- 1..1 LG1 1 /zfs LG1 node is connected to the same single network switch. Load Generator Configuration Notes ---------------------------------- The load generator client was mounted against the single filesystem on all FXT nodes. Uniform Access Rule Compliance ============================== The load-generating client hosted 768 processes. The assignment of 768 processes to 3 network interfaces was done such that they were evenly divided across all network paths to the FXT appliances. The filesystem data was evenly distributed across all disks and FXT appliances and Core filer storage servers. Other Notes =========== N/A ================================================================================ Generated on Mon Apr 07 08:54:58 2014 by SPECsfs2008 ASCII Formatter Copyright (C) 1997-2008 Standard Performance Evaluation Corporation