hiyohiyo / diskspd

DiskSpd

DiskSpd is a storage performance tool from the Windows, Windows Server and Cloud Server Infrastructure engineering teams at Microsoft. Please visit https://github.com/Microsoft/diskspd/wiki for updated documentation. The latest update to DiskSpd can be downloaded from https://aka.ms/diskspd.

In addition to the tool itself, this repository hosts measurement frameworks which utilize DiskSpd. The initial example is VM Fleet that was used for the Windows Server 2016 Hyper-Converged Storage Spaces Direct work.

This project has adopted the Microsoft Open Source Code of Conduct. For more information see the Code of Conduct FAQ or contact [e-mail] with any additional questions or comments.

What's New?

DISKSPD

DISKSPD 2.0.20a 2/28/2018

• Changes that may require rebaselining of results:
  • New random number generator that may show an observable decreased cost
  • Switched to 512-byte aligned buffers with the -Z option to increase performance
• New -O option for specifying the number of outstanding IO requests per thread
• New -Zr option for per-IO randomization of write buffer content
• XML: Adds a new <ThreadTarget> element to support target weighting schemes
• Enhanced statistics captured from IOPS data
• Added support for validating XML profiles using an in-built XSD
• Added support for handling RAW volumes
• Updated CPU statistics to work on > 64-core systems
• Updated calculation and accuracy of CPU statistics
• Re-enable support for ETW statistics

DISKSPD 2.0.18a 5/31/2016

• update /? example to use -Sh v. deprecated -h
• fix operation on volumes on GPT partitioned media (:)
• fix IO priority hint to proper stack alignment (if not 8 byte, will fail)
• use iB notation to clarify that text result output is in 2^n units (KiB/MiB/GiB)

DISKSPD 2.0.17a 5/01/2016

• -S is expanded to control write-through independent of OS/software cache. Among other things, this allows buffered write-through to be specified (-Sbw).
• XML: adds a new <WriteThrough> element to specify write-through
• XML: <DisableAllCache> is no longer emitted (still parsed, though), in favor or <WriteThrough> and <DisableOSCache>
• Text output: OS/software cache and write-through state are now documented separately (adjacent lines)
• Latency histogram now reports to 9-nines (one part in one billion) in both text and XML output
• Error message added for failure to open write-content source file (-Z<size>,<file>)

VM Fleet

VM Fleet 0.9 10/2017 (minor)

• watch-cpu: now provides total normalized cpu utility (accounting for turbo/speedstep)
• sweep-cputarget: now provides average CSV FS read/write latency in the csv

VM Fleet 0.8 6/2017

• get-cluspc: add SMB Client/Server and SMB Direct (not defaulted in Storage group yet)
• test-clusterhealth: flush output pipeline for Debug-StorageSubsystem output
• watch-cluster: restart immediately if all child jobs are no longer running
• watch-cpu: new, visualizer for CPU core utilization distributions

VM Fleet 0.7 3/2017

• create/destroy-vmfleet & update-csv: don't rely on the csv name containing the friendlyname of the vd
• create-vmfleet: err if basevhd inaccessible
• create-vmfleet: simplify call-throughs using $using: syntax
• create-vmfleet: change vhdx layout to match scvmm behavior of seperate directory per VM (important for ReFS MRV)
• create-vmfleet: use A1 VM size by default (1VCPU 1.75GiB RAM)
• start-vmfleet: try starting "failed" vms, usually works
• set-vmfleet: add support for -SizeSpec for A/D/D2v1 & v2 size specification, for ease of reconfig
• stop-vmfleet: pass in full namelist to allow best-case internal parallelization of shutdown
• sweep-cputarget: use %Processor Performance to rescale utilization and account for Turbo effects
• test-clusterhealth: support cleaning out dumps/triage material to simplify ongoing monitoring (assume they're already gathered/etc.)
• test-clusterhealth: additional triage output for storport unresponsive device events
• test-clusterhealth: additional triage comments on SMB client connectivity events
• test-clusterhealth: new test for Mellanox CX3/CX4 error counters that diagnose fabric issues (bad cable/transceiver/roce specifics/etc.)
• get-log: new triage log gatherer for all hv/clustering/smb event channels
• get-cluspc: new cross-cluster performance counter gatherer
• remove run-<>.ps1 scripts that were replaced with run-demo-<>.ps1
• check-outlier: EXPERIMENTAL way to ferret out outlier devices in the cluster, using average sampled latency

VM Fleet 0.6 7/18/2016

• CPU Target Sweep: a sweep script using StorageQoS and a linear CPU/IOPS model to build an empirical sweep of IOPS as a function of CPU, initially for the three classic small IOPS mixes (100r, 90:10 and 70:30 4K). Includes an analysis script which provides the linear model for each off of the results.
• Update sweep mechanics which allow generalized specification of DISKSPD sweep parameters and host performance counter capture.
• install-vmfleet to automate placement after CSV/VD structure is in place (add path, create dirs, copyin, pause)
• add non-linearity detection to analyze-cputarget
• get-linfit is now a utility script (produces objects describing fits)
• all flag files (pause/go/done) pushed down to control\flag directory
• demo scripting works again and autofills vm/node counts
• watch-cluster handles downed/recovered nodes gracefully
• update-csv now handles node names which are logical prefixes of another (node1, node10)

Source Code

The source code for DiskSpd is hosted on GitHub at:

https://github.com/Microsoft/diskspd

Any issues with DiskSpd can be reported using the following link:

https://github.com/Microsoft/diskspd/issues