SPEC SFS®2014_swbuild Result

Copyright © 2014-2017 Standard Performance Evaluation Corporation

WekaIO SPEC SFS2014_swbuild = 500 Builds
WekaIO Matrix Overall Response Time = 3.06 msec


Performance

Business
Metric
(Builds)
Average
Latency
(msec)
Builds
Ops/Sec
Builds
MB/Sec
500.925001397
1001.150004794
1501.4750061191
2001.71000071588
2503.41250091986
3002.51500112383
3503.01750122780
4004.02000133177
4505.72250123574
5008.82488393953
Performance Graph


Product and Test Information

WekaIO Matrix
Tested byWekaIO
Hardware AvailableNot Applicable
Software AvailableApril 2017
Date TestedJune 2017
License Number4553
Licensee LocationsSan Jose, California

WekaIO Matrix Software solves the challenge of scaling high performance file workloads from tens to thousands of server clients. Matrix Software transforms a cluster of compute servers (on-premises or in AWS) into a globally shared, POSIX compliant file system that delivers industry leading performance and scale for enterprise file workloads and HPC environments. Matrix Software makes it simple to install and manage hundreds of compute nodes from a single management console. For peak performance needs, additional Matrix instances can be dynamically added to scale performance, while capacity can be scaled independently within the compute nodes. WekaIO Matrix is flash native providing the highest performance to applications, and its integrated tiering mechanism automatically moves cold data to any S3 or Swift compliant object store (on-premises or public cloud), while maintaining data locality to the applications. All data is presented as a single global namespace.

WekaIO Matrix is ideal for technical computing workloads such as EDA design, Big Data analytics, software development, genomic sequencing, financial analytics, media rendering and HPC research. The software can be deployed natively in a hyperconverged mode with zero additional footprint, or as a dedicated storage appliance. Because it is a software only solution it is up to 80% less costly than external flash based NAS appliances. This test platform is a hyperconverged implementation.

Solution Under Test Bill of Materials

Item NoQtyTypeVendorModel/NameDescription
160AWS EC2 InstanceAmazonr3.8xlarge instanceAmazon r3.8 instances are optimized for memory-intensive applications and offer lower price per GiB of RAM. Features: High Frequency Intel Xeon E5-2670 v2 (Ivy Bridge) Processors, 640GB of SSD Storage, Support for Enhanced Networking

Configuration Diagrams

  1. Solution Under Test

Component Software

Item NoComponentTypeName and VersionDescription
1Hyperconverged Server NodesMatrixFS File System2.5.2WekaIO Matrix is a distributed and parallel POSIX file system that runs on any SSD enabled commodity server or cloud compute instance and forms a single cluster. The file system presents a single high performance, scalable global namespace to the applications running in the cluster.
2Hyperconverged Server NodesOperating SystemCENTOS 6.7The operating system on the hyperconverged server was 64-bit CENTOS Version 6.7.

Hardware Configuration and Tuning - Virtual

SR-IOV
Parameter NameValueDescription
SR-IOVEnabledEnables CPU virtualization technology

Hardware Configuration and Tuning Notes

SR-IOV was enabled in the EC2 instance BIOS. No additional hardware tuning was required.

Software Configuration and Tuning - Virtual

Hyperconverged Server Nodes
Parameter NameValueDescription
Jumbo Frames9000Enables up to 9000 bytes of Ethernet Frames

Software Configuration and Tuning Notes

The MTU is required and valid for all environments and workloads.

Service SLA Notes

Not applicable

Storage and Filesystems

DescriptionData ProtectionStable StorageQtyUsable GiB
MatrixFS Enabled Amazon EC2 r3.8xlarge instances used as shared file system15+2Yes135.2 TiB
Total35.2 TiB
Number of Filesystems1
Total Capacity24176 GiB
Filesystem TypeMatrixFS

Filesystem Creation Notes

A single WekaIO Matrix file system was created and distributed evenly across all 60 instances in the Amazon cluster. Data was protected to an 15+2 failure level. All data and metadata is distributed evenly across the cluster.

Storage and Filesystem Notes

WekaIO MatrixFS was created and distributed evenly across all 60 instances in the Amazon cluster. The deployment model is as a hyperconverged architecture where both storage and application services are run on the same infrastructure. Each EC2 instance has 32 vCPU, 244GiB of memory and 2x320GB (total 640GB) of SSD local to the instance. Two cores of the available 32 (6.25%) were provisioned with MatrixFS, the remaining 30 cores were available for application usage. Each instance had 13GB (5%) of memory assigned to WekaIO.

Transport Configuration - Virtual

Item NoTransport TypeNumber of Ports UsedNotes
110GbE Virtual NIC60Interface Notes

Transport Configuration Notes

WekaIO Matrix utilizes the shared 10GbE Virtual NIC on each r3.8xlarge instance. 1x10GbE Virtual NIC per EC2 instance for a total of 60. Applications and storage run on the same network.

Switches - Virtual

Item NoSwitch NameSwitch TypeTotal Port CountUsed Port CountNotes
1AWS10GbE with Enhanced NetworkingAWSAWSSwitch Notes

Processing Elements - Virtual

Item NoQtyTypeLocationDescriptionProcessing Function
1120vCPU coreAWS EC2 r3.8xlarge instanceIntel(R) Xeon(R) CPU E5-2670 v2 @ 2.50GHzProcessing Function

Processing Element Notes

Each amazon instance had 32 vCPUs. WekaIO used only 2 cores per instance.

Memory - Virtual

DescriptionSize in GiBNumber of InstancesNonvolatileTotal GiB
AWS EC2 r3.8xlarge instance memory24460V14640
Grand Total Memory Gibibytes14640

Memory Notes

WekaIO utilized 13GBytes of memory per instance

Stable Storage

Each WekaIO instance used the EC2 instance ephemeral drive for stable storage. Data was protected using WekaIO Matrix Distributed Data Protection (15+2). The data is journaled as well as UPS protected accross multiple layers.

Solution Under Test Configuration Notes

The solution under test was a standard WekaIO Matrix enabled cluster in hyperconverged 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. The storage services and client applications share the same instances without the need for any additional dedicated storage instances.

Other Solution Notes

None

Dataflow

The EC2 instances are both the load generators and the storage nodes. All of the instances accessed the same WekaIO Matrix shared file system that was distributed and protected across all of the EC2 instances.

Other Notes

None

Other Report Notes

None


Generated on Wed Jul 5 15:24:17 2017 by SpecReport
Copyright © 2014-2017 Standard Performance Evaluation Corporation