VMware Cloud Community
christianZ
Champion
Champion

New !! Open unofficial storage performance thread

Hello everybody,

the old thread seems to be sooooo looooong - therefore I decided (after a discussion with our moderator oreeh - thanks Oliver -) to start a new thread here.

Oliver will make a few links between the old and the new one and then he will close the old thread.

Thanks for joining in.

Reg

Christian

574 Replies
larstr
Champion
Champion

merkproctor wrote:

Anybody else testing a Flash Array?  I'm not seeing anyone posting anything close to this.  Thx

We should probably use a different test for All Flash Arrays (AFA). These Arrays often have inline dedupe and compression, some of them get decreased performance when disks fill up above a certain level and many of them will suffer from less performance after a test has been running for a few hours.

With the test files provided for iometer in this thread the tested data sets are too small and for many AFAs will only let the test run against a memory cache. The data that iometer inputs in the data sets are not random and inline dedupe may affect the results.

A newer version of IOmeter supports randomized data files and we should probably create a new thread with a new test set. I'm however not sure a generic test like this will be enough. The problem is that a test should run through all active paths agains the AFA and would maybe require multiple VMs on multiple hosts.

Lars

Reply
0 Kudos
Mike_b79
Contributor
Contributor

SERVER TYPE: Windows 2012 R2

CPU TYPE / NUMBER: Xeon E5-2609v3

HOST TYPE: Cisco UCS B200 M4 blade

STORAGE TYPE / DISK NUMBER / RAID LEVEL: Nimble CS210 4X80GB SSD, 8X1TB HDD

Test name

LatencyAvg iopsAvg MBpscpu load
Max Throughput-100%Read8.49695221714%
RealLife-60%Rand-65%Read5.58101047818%
Max Throughput-50%Read6.81795224819%
Random-8k-70%Read5.20109168519%
Reply
0 Kudos
Mike_b79
Contributor
Contributor

SERVER TYPE: Windows 2012 R2

CPU TYPE / NUMBER: Xeon E5-2609v3

HOST TYPE: Cisco UCS B200 M4 blade

STORAGE TYPE / DISK NUMBER / RAID LEVEL: EMC VNX 5200, 2x100GB SSD, 5x200GB SSD, 15x600GB SAS, 8x2TB NL

Test name

LatencyAvg iopsAvg MBpscpu load
Max Throughput-100%Read3.181853557927%
RealLife-60%Rand-65%Read4.54105808227%
Max Throughput-50%Read5.081122935019%
Random-8k-70%Read3.86104538134%
Reply
0 Kudos
pbhite
Enthusiast
Enthusiast

SERVER TYPE: 2012 R2

CPU TYPE / NUMBER: E5620 x 2

HOST TYPE: Dell M610

STORAGE TYPE / DISK NUMBER / RAID LEVEL: Compellent SC8000 / 24x 200GB SLC SSD (RAID 10), 12x 1.6TB MLC SSD (RAID 5), 80x 4TB 7.2k (RAID 6)

pastedImage_0.png

Reply
0 Kudos
pbhite
Enthusiast
Enthusiast

SERVER TYPE: 2012 R2

CPU TYPE / NUMBER: E5620 x 2

HOST TYPE: Dell M610

STORAGE TYPE / DISK NUMBER / RAID LEVEL: Compellent Series 40 / 68x 500GB 10k SAS (RAID 10, RAID5) 84x 2TB 7.2k (RAID 10/RAID 6)

Test name

LatencyAvg iopsAvg MBpscpu load
Max Throughput-100%Read6.18103513230%
RealLife-60%Rand-65%Read5.5611512890%
Max Throughput-50%Read7.9780302500%
Random-8k-70%Read5.6011420890%
Reply
0 Kudos
mbertin
Contributor
Contributor

SERVER TYPE:VM 2012 R2
CPU TYPE / NUMBER: Xeon e5-2620 *2 (vm 2vcpu)
HOST TYPE: esxi6 (dl380 g9)
STORAGE TYPE / DISK NUMBER / RAID LEVEL: dual datacore HP dl380 g9 SSV10 psp4U1 with DAS HP d3700 (smart array p810 4go) 26 disk raid 5 / fc 4gb

Test name

Latency

Avg iops

Avg MBps

cpu load

Max Throughput-100%Read

4.68

12866

402

6%

RealLife-60%Rand-65%Read

1.00

36606

285

18%

Max Throughput-50%Read

2.50

23547

735

10%

Random-8k-70%Read

0.98

34500

269

20%

Reply
0 Kudos
lukejf
Contributor
Contributor

hi Guys

Just wondering if anyone has the .icf file that will work with the io analyser ovf template. I have managed to import what i thought is the correct icf file however it does not report back to the web console. Seems to work ok if i run it manually

Reply
0 Kudos
lukejf
Contributor
Contributor

Test nameLatencyAvg iopsAvg MBpscpu load
Max Throughput-100%Read0.5074918234160%
RealLife-60%Rand-65%Read17.6716331261%
Max Throughput-50%Read0.2739433123225%
Random-8k-70%Read17.0820631638%

Old style VMTN communities table:

SERVER TYPE: Lenovo x3650M5 CPU TYPE / NUMBER: Dual Xeon E5-2630V3 HOST TYPE: ESXI 5.5 STORAGE TYPE / DISK NUMBER / RAID LEVEL: Local Disk M5210 8 x 1.8TB 12GbSAS 10K Raid 10  |*TEST NAME*|*Avg Resp. Time ms*|*Avg IOs/sec*|*Avg MB/sec*|*% cpu load*| |*Max Throughput-100%Read*|0.50|74918|2341|60%| |*RealLife-60%Rand-65%Read*|17.67|1633|12|61%| |*Max Throughput-50%Read*|0.27|39433|1232|25%| |*Random-8k-70%Read*|17.08|2063|16|38%| 
Reply
0 Kudos
mpogr
Enthusiast
Enthusiast

SERVER TYPE: Windows 10 Enterprise VM, 4GB vRAM
CPU TYPE / 4 vCPUs
HOST TYPE: 2xAMD Opteron 6344 CPU (12 cores each), 64GB RAM 
STORAGE TYPE / DISK NUMBER / RAID LEVEL: Self-built NAS on Intel E3-1270 v2 Xeon CPU, 32GB RAM running 
ZFS on Linux (CentOS 7.2) RAID10 array of 6xHDD WD RE 4TB, Intel SSD S3700 100GB as SLOG and Intel 335 240GB as LARC 
via direct 40Gb Infiniband link (directly connected Mellanox Connext-X2 adapters) 
Test nameLatencyAvg iopsAvg MBpscpu load
Max Throughput-100%Read1.1432322101014%
RealLife-60%Rand-65%Read0.90311792438%
Max Throughput-50%Read1.242900790611%
Random-8k-70%Read0.923157824615%

Old style VMTN communities table:

SERVER TYPE: Windows 10 Enterprise VM, 4GB vRAM
CPU TYPE / 4 vCPUs
HOST TYPE: 2xAMD Opteron 6344 CPU (12 cores each), 64GB RAM 
STORAGE TYPE / DISK NUMBER / RAID LEVEL: Self-built NAS on Intel E3-1270 v2 Xeon CPU, 32GB RAM running 
ZFS on Linux (CentOS 7.2) RAID10 array of 6xHDD WD RE 4TB, Intel SSD S3700 100GB as SLOG and Intel 335 240GB as LARC 
via direct 40Gb Infiniband link (directly connected Mellanox Connext-X2 adapters) 

|*TEST NAME*|*Avg Resp. Time ms*|*Avg IOs/sec*|*Avg MB/sec*|*% cpu load*|
|*Max Throughput-100%Read*|1.14|32322|1010|14%|
|*RealLife-60%Rand-65%Read*|0.90|31179|243|8%|
|*Max Throughput-50%Read*|1.24|29007|906|11%|
|*Random-8k-70%Read*|0.92|31578|246|15%|
Reply
0 Kudos
datacoredcie
Contributor
Contributor

Unbenannt.jpg

Hello together,

we are implemented a Datacore HyperConverged Virtual SAN with two identical Fujitsu RX 2540 M2 with 256GB RAM + 2*12Core CPUs (48 Logical CPUs) + 2*300GB SAS for Local DCS Installation and 2*400GB SLC SSD + 12*600GB SAS HDD for VMWare Datastore. We have 2*10GB NICs with each 2 Ports each Server for 2*Mirror and 2*FrontEnd iSCSI Ports. Each Datacore Server have 40 GB RAM Cache

Hypervisor is an VMWare ESXi 6.5 the lastest Version an Datacore SanSymphony V 10 PSP6 Update 2

SSVHCV.jpg

Reply
0 Kudos
marshallmathers
Enthusiast
Enthusiast

SERVER TYPE: VM (Windows Server 2012 R2)

CPU TYPE / NUMBER: 4 vCPU

HOST TYPE: DELL R620 (VMware 5.5 U3)

STORAGE TYPE / DISK NUMBER / RAID LEVEL: IBM V3700 10 GbE iSCSI, 16 x 600GB 10k SAS / 6 x 800GB SSD / RAID 5

Test nameLatencyAvg iopsAvg MBpscpu load
Max Throughput-100%Read8.4736161130%
RealLife-60%Rand-65%Read8.796365490%
Max Throughput-50%Read9.5350841580%
Random-8k-70%Read8.916432500%

something def up with my setup. should be running at 10 GIG . Any suggestions ?

Reply
0 Kudos
marshallmathers
Enthusiast
Enthusiast

Well now I am getting this after unchecked "Test connection Rate " in IOMETER (previously was set to 500 transactions per connection)

Test nameLatencyAvg iopsAvg MBpscpu load
Max Throughput-100%Read7.1988992780%
RealLife-60%Rand-65%Read8.487550580%
Max Throughput-50%Read10.0763521980%
Random-8k-70%Read8.247767600%
Reply
0 Kudos
kesslerseb
Contributor
Contributor

SERVER TYPE: Server 2016, VMWare 6.0

CPU TYPE / NUMBER: 4x vCPU, 4GB Ram, 60GB VMDK

HOST TYPE: Lenovo System X 3500, 2x Xeon 2640, 256GB RAM, 12G SAS

STORAGE TYPE / DISK NUMBER / RAID LEVEL: V3700V2, 11x 1200GB SAS RAID 5, 4x400GB SSD RAID 5

SAN_V3700.PNG

Reply
0 Kudos
nizakia
Contributor
Contributor

hi,

Can these values displayed as output after copying csv file in vmktree/iometer site becahnged to display more than these four fields like latency,Avg IOPs, Avg mbps and cpu load?

thanks

Reply
0 Kudos
larstr
Champion
Champion

Yes, it is possible. What values would you like to see?

Reply
0 Kudos