SPECsfs2008_nfs.v3 Result

Throughput (ops/sec) Response (msec) 76014 0.6 152106 0.8 228368 0.9 304812 1.1 381040 1.3 457435 1.4 534709 1.7 610904 2.2 688765 3.0 765667 5.0

Tested By	NetApp, Inc.
Product Name	Data ONTAP 8.1 Cluster-Mode (12-node FAS6240)
Hardware Available	Nov 2010
Software Available	Sept 2011
Date Tested	Sept 2011
SFS License Number	33
Licensee Locations	Sunnyvale, CA USA

Data ONTAP 8.1 Cluster-Mode extends the storage domain of a high-availability (HA) pair of controllers to multiple pairs of controllers. With Data ONTAP 8.1 Cluster-Mode, capacity scales from terabytes to tens of petabytes, all transparent to your running applications. Your storage is virtualized across as many as 24 nodes, managed as a single logical pool of resources and name space. Virtualizing your storage across multiple pairs of controllers provides nearly limitless scalability for even the most data-intensive environments, regardless of network protocol, SAN or NAS. Data ONTAP 8.1 offers the first massively scalable unified storage platform with support for FC, iSCSI, FCoE, NFS and CIFS.

Item No	Qty	Type	Vendor	Model/Name	Description
1	12	Storage Controller	NetApp	FAS6240	FAS6240 with SAS CTRL and IO Expander
2	12	4-port SAS IO card	NetApp	IO Board SAS 4-port X2067-R6	4-port SAS IO Card
3	36	Disk Shelves with SAS Disk Drives	NetApp	DS4243-1511-24S-R5-C	DS4243 disk shelf with 24x450GB, 15K, SAS HDD
4	12	Flash Cache Module	NetApp	PAMII-512-R5	Flash Cache Module 512GB
5	1	Software license	NetApp	SW-6240-Cluster License	Data ONTAP 8.1 Cluster mode License
6	1	Software License	NetApp	SW-6240-NFS	NFS Software license SW-6240-NFS
7	2	Network Switch	Cisco	Cisco Nexus 5020	Cisco Nexus 5020 switches

OS Name and Version	Data ONTAP 8.1 Cluster-Mode
Other Software	None
Filesystem Software	Data ONTAP 8.1 Cluster-Mode

Name	Value	Description
vol modify <volume_name> -atime-update	false	Disable access time updates (applied to all volumes)

Server Tuning Notes

N/A

Description	Number of Disks	Usable Size
450GB SAS 15K RPM Disk Drives	864	287.0 TB
Total	864	287.0 TB

Number of Filesystems	single namespace
Total Exported Capacity	144 TB
Filesystem Type	WAFL
Filesystem Creation Options	The file-system was created using default values. An export policy was created giving full-access and applied to the file system.
Filesystem Config	48 RAID-DP(Double Parity) groups of 17 disks each were created across all the disks.
Fileset Size	89059.3 GB

The storage configuration consisted of 12 nodes connected in 6 SFO (storage failover) pairs forming a single cluster. Each node was connected to its own and partner's disks. A single disk pool or "aggregate" was created on each node comprised of 4 RAID-DP raid groups, each composed of 15 data disks and 2 parity disks which held the data for the file-system. A separate aggregate consisting of 3 disks in a single RAID-DP group was created to hold the Data ONTAP operating system files. Each node was allocated a spare disk.A single virtual server or "vserver" was created on the cluster, spanning all physical nodes. A flexible volume was then created on the data aggregate of each node in the cluster. Each volume was junctioned at the root of the single namespace of the vserver. Each volume was striped across all the disks in the data aggregate.

Item No	Network Type	Number of Ports Used	Notes
1	Jumbo frame 10 Gigabit Ethernet	36	24 ports (2 per node) were used for the cluster network and 12 ports (1 per node) were used for the data network.

Network Configuration Notes

The two cluster ports from each node were connected to a pair of Cisco Nexus 5020 switches, one port to each switch to provision the cluster network. This provides high availability for the cluster network in case of port or link failure. One port from each node and each load generator was connected to a Cisco Nexus 5596 switch for the data network. The data and cluster networks were on separate subnets. All the ports (cluster and data) were configured to use jumbo frames.

Each load generator was connected to the data network switch via a single 10GbE port. MTU size of 9000 was set for all connections to the switch.

Item No	Qty	Type	Description	Processing Function
1	24	CPU	2.53GHz Intel Xeon(tm) Processor E5540	Networking, NFS protocol, WAFL filesystem, RAID/Storage drivers, Clustering

Processing Element Notes

Each storage controller has two physical processors and each physical processor is made up of four cores.

Description	Size in GB	Number of Instances	Total GB	Nonvolatile
Storage Controller Main Memory	48	12	576	V
NVRAM Non-volatile Memory on PCIe adapter	4	12	48	NV
Flash Cache Module memory	512	12	6144	V
Grand Total Memory Gigabytes			6768

Memory Notes

Each storage controller has main memory that is used for the operating system and caching filesystem data. The FlashCache module is a read cache used for caching filesystem data. A separate, integrated battery-backed RAM module is used to provide stable storage for writes that may not have been written to disk.

The WAFL filesystem logs writes and other filesystem data modifying transactions to the NVRAM adapter. In a storage-failover configuration, as in the system under test, such transactions are also logged to the NVRAM on the partner storage controller so that, in the event of a storage controller failure, any transactions on the failed controller can be completed by the partner controller. Filesystem modifying CIFS/NFS operations are not acknowledged until after the storage system has confirmed that the related data are stored in NVRAM adapters of both storage controllers (when both controllers are active). The battery-backed NVRAM ensures that any uncommitted transactions are preserved for at least 72 hours. In addition, de-staging to Flash on NVRAM preserves these transactions permanently.

The system under test consisted of 12 FAS6240 storage controllers and 36 storage shelves, each with 24 450GB SAS disk drives. The controllers were running Data ONTAP 8.1 software operating in Cluster-Mode. The 12 storage controllers(nodes) were configured in a storage failover (SFO) configuration and connected to their respective disk shelves in a multi-path high-availability (MPHA) configuration. The SFO was provided by the storage failover software option in conjunction with an InfiniBand interconnect provided on the NVRAM adapter. Each node has physical resources that can be shared across the cluster providing a single namespace. The cluster network (N1) is comprised of 2 Cisco Nexus 5020 switches to provide redundancy. All the nodes in the cluster are connected through the cluster network and all the data throughout the cluster is accessible from any node. The data network (N2) is comprised of a Cisco Nexus 5596 switch. Each node and load-generator have a single 10GbE port connected to this switch. The data and cluster networks are on separate subnets. All ports and interfaces on the data and cluster networks have Jumbo frames enabled.

All standard data protection features, including background RAID and media error scrubbing, software validated RAID checksum, and double disk failure protection via double parity RAID (RAID-DP) were enabled during the test.

Item No	Qty	Vendor	Model/Name	Description
1	18	Fujitsu	Fujitsu Primergy RX300 S6	Fujitsu RX300-S6 rack servers with 48GB memory running RHEL 5.5
2	1	Cisco	Nexus 5596	Cisco Nexus 5596UP Switch

LG Type Name	LG1
BOM Item #	1
Processor Name	Intel Xeon E5645
Processor Speed	2.40GHz
Number of Processors (chips)	2
Number of Cores/Chip	6
Memory Size	48 GB
Operating System	RHEL5.5 Kernel 2.6.18-194.el5
Network Type	10 Gigabit Ethernet

Benchmark Parameters

Network Attached Storage Type	NFS V3
Number of Load Generators	18
Number of Processes per LG	240
Biod Max Read Setting	2
Biod Max Write Setting	2
Block Size	AUTO

Testbed Configuration

LG No	LG Type	Network	Target Filesystems	Notes
1..18	LG1	Data Network (N2)	Single Namespace containing volumes vol1 to vol12.	See UAR Notes

Load Generator Configuration Notes

All clients accessed all mount points. Mount points were assigned to each client in a uniform manner making sure the load is evenly distributed across all clients and all interfaces.

For UAR compliance, each flexible volume was mapped to a subdirectory of the global namespace under root (/data1,/data2,...,/data12). Each volume was accessed over all data network interfaces (ip addresses ip1...ip12) such that each volume had a unique mount point on each node. This ensured that 1/12th of the total access on each node was done through a local interface and 11/12th of the total access was done through interfaces which were remote to the node the volume resides on. For these remote accesses, data traversed the cluster network. There were a total of 12 IP addresses available for data access, one per node. Each client mounted all 12 volumes using all 12 target IP addresses, cycling through the IP addresses first for a particular volume and then moving to the next volume after using all IP addresses. Since the total required mount points per client were 240 and the above rotation gave 144 mount points, we circled back to the start of the series and continued till 240 mount points. Each successive client continued from where the previous client left off. For example LG1 was assigned the following mount-point list: ip1:/data1, ip2:/data1,...,ip12:/data1,ip1:/data2,...,ip12:/data2,...,ip1:/data12,...,ip12:/data12,ip1:/data1,..,ip12:/data1,..,ip1:/data8,..,ip12:/data8. LG2 continued from where LG1 left off:ip1:/data9,ip2:/data9,..,ip12:/data9,..,ip12:/data1,..,ip1:/data4,..,ip12:/data4. This was repeated for all 18 clients. This ensured that data access to every volume was uniformly distributed across all clients and target IP addresses. The volumes were striped evenly across all the disks in the aggregate using all the SAS adapters connected to storage backend.

Other test notes: None.

NetApp is a registered trademark and "Data ONTAP", "FlexVol", and "WAFL" are trademarks of NetApp, Inc. in the United States and other countries. All other trademarks belong to their respective owners and should be treated as such.

• Config Diagram