Re: How to get StorageAmount used by Thin provisioned LUNs through SMI-S

[Paul von Behren]

The SpaceConsumed  property of the AlloatedromStoragePool association(s) referencing the StorageVolume (LUN)

On Thu, Mar 13, 2014 at 3:18 PM, Aditya Deo <adit...@gmail.com> wrote:

Hi,

I am trying to get StorageAmount used by Thin provisioned LUNs through SMI-S API calls. How can I get this data?

 

Thanks,

Aditya

--
You received this message because you are subscribed to the Google Groups "SMI-S Developers Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to smi-s-developers-...@googlegroups.com.
To post to this group, send email to smi-s-devel...@googlegroups.com.
Visit this group at http://groups.google.com/group/smi-s-developers-group.
For more options, visit https://groups.google.com/d/optout.

[Aditya]

Hi paulvb, 

Thanks for the help. I am new to all this and struggling a bit. I got an instance of Volume(thin provisioned) and followed Clar_AllocatedFromStoragePoolView_SPV_SV association(we are using vnx clariion).

In the instance block, I got 

Instance of Clar_StoragePoolView {

uint64 AFSPSpaceConsumed = 2954451353088;

string SCCElementName = U+VNXPoolForBlock1;

.

.

}

The AFSPSpaceConsumed is not actually making sense to me, as the real Used value is 3.006 GB for that LUN. what can be possibly wrong?

Also, how can I get IOPS for pools and is there any way to get Latency value?

I really appreciate your help. Thanks in advance.

-

Thanks,

Aditya

On Thursday, March 13, 2014 11:17:43 PM UTC-4, paulvb wrote:

The SpaceConsumed  property of the AlloatedromStoragePool association(s) referencing the StorageVolume (LUN)

On Thu, Mar 13, 2014 at 3:18 PM, Aditya Deo wrote:

Hi,

I am trying to get StorageAmount used by Thin provisioned LUNs through SMI-S API calls. How can I get this data?

 

Thanks,

Aditya

--
You received this message because you are subscribed to the Google Groups "SMI-S Developers Group" group.

To unsubscribe from this group and stop receiving emails from it, send an email to smi-s-developers-group+unsub...@googlegroups.com.

[Paul von Behren]

The storage amount used in a thin provisioned LUN may change every time there is a write from the file system  (or database) using the LUN.  So minor changes like the difference you reported may occur within a second on a live LUN.  

Another thing to consider is whether the 3.006G you reported in gigabytes or Gibibytes?  It would be common for UI SW to report in gigabytes, but in this case, the difference is even larger (3.006 gigabytes is 2.607289384391928 gibibytes).

As for  IOPS and Latency; there is an SMI-S Block Serer Performance profile with some performance info.  But I think you might be looking some something different.  Most of the time when people talk about storage latency, tjey are talking about latency in the entire round trip, starting at the HBA (or better yet, the host SW).  Array metrics can't possibly provide information on the latency outside the array.  But you can get the latency using standard OS tools on the system(s) using the LUN. On Linux or UNIX systems, you can use iostats.  Windows has a lat of tools including the COntrol Panel snapins and powershell/WMI tools. Once you have acquired latency info, you also have what you need to determine IOPS.

Note that EMC has a nice doc on their SMIS support: 

https://developer-content.emc.com/developer/devcenters/storage/snia/smi-s/downloads/SMIProvider_V4.6PG.pdf

To unsubscribe from this group and stop receiving emails from it, send an email to smi-s-developers-...@googlegroups.com.

[Aditya]

Hi Paul,

Thanks for the detailed explanation. When I went through different instances of thin provisioned volumes and used Clar_AllocatedFromStoragePoolView_SPV_SV relation, I get back Pool instances and not the Volume instances. 
Also could it be possible to have your contact information, so that I can ask more VNX related questions? 

I really appreciate your help.

Thanks,

Aditya

Hi paulvb, 

To unsubscribe from this group and stop receiving emails from it, send an email to smi-s-developers-group+unsubscri...@googlegroups.com.

To post to this group, send email to smi-s-devel...@googlegroups.com.
Visit this group at http://groups.google.com/group/smi-s-developers-group.
For more options, visit https://groups.google.com/d/optout.

[Paul von Behren]

Sorry for the slow response; I've been attending a conference with flaky network access.

AllocatedFromStoragePool is used two ways.  It can associate two StoragePools or  associate one StoragePool and one StorageVolume. Also note that some StoragePools will not have any associated StorageVolumes.  So the easiest approach is to start with a StorageVolume instance. Use a References CIM client request and the association class  CIM_AllocatedFromStoragePool, specifying the role as Dependent.  I see EMC documentation describing this approach (which conforms to SMI-S Block Services) for CLARiiON systems.  You included results from AllocatedFromStoragePoolView.  The same pattern applies - may be bewteen two pools or between a pool or volume.  Note that the view classes are optional; EMC may not support them for volumes.  I can't tell from the partial example you provided whether it references a volume.  But I recommend using the non-view AllocatedFromStoragePool class - since it is required.

I recall you mentioned VNX. I believe the VNX systems support both array and NAS behavior.  If capacity is allocated for NAS, you might see LogicalDisk instances rather than StorageVolume referenced by AllocatedFromStoragePool.  But I don't think the SMI-S NAS profiles requires use of block services and the relationship between filesystems and pools is vendor specific.  I'm less familiar with the NAS profiles; please check the SMi-S specs (or better yet, EMC documents) if you looking at NAS instances.

Paul

On Mon, Mar 17, 2014 at 4:27 PM, Aditya <adit...@gmail.com> wrote:

Hi Paul,

Thanks for the detailed explanation. I tried traversing Clar_AllocatedFromStoragePoolView_SPV_SV relation on different thin LUNs, but received pool instance. How can I get back Volume instances with spaceConsumed in it. 

I really appreciate your help but I'm not able to figure out correct relations.

Thanks,

Aditya 

 

--
You received this message because you are subscribed to the Google Groups "SMI-S Developers Group" group.

[Aditya]

Hi Paul,

Thanks for the reply. In the meantime, I figured out he relation to get space consumed by thin LUNs. I get back instance of LUN and using clar_baseInstance_SV_VV association, we get back AFSPSpaceConsumed attribute which reflects the used value.

Your posts/replies are very helpful and helping a lot in find right solution. 

Thanks,

Aditya

To unsubscribe from this group and stop receiving emails from it, send an email to smi-s-developers-group+unsub...@googlegroups.com.

[Aditya]

Hi Paul,

As mentioned in earlier thread,  I am trying to get IOPS between storagePool and and LUNs. After traversing "Clar_ConcreteDependency_DD_USP" association from poolInstance, I got back following data 

instance of Clar_DiskDrive {

  Caption = "NL_SAS";

  CreationClassName = "Clar_DiskDrive";

  DefaultBlockSize = 512;

  Description = "Manufacturer=SEAGATE SerialNumber=Z1N4KQW6";

  DeviceID = "CLARiiON+0_0_10";

  DiskType = 2;

  ElementName = "CLARiiON Disk 0_0_10";

  EMCActualBlocks = 0;

  EMCDiskGroupNumber = 4294967295;

  EMCFreeBlocks = 1923469313;

  EMCInUse = true;

  EMCIsVault = false;

  EMCLocality = 2;

  EMCManufacturer = "SEAGATE";

  EMCNumberOfBlocks = 1923469312;

  EMCSerialNumber = "Z1N4KQW6";

  EnabledDefault = 2;

  EnabledState = 5;

  Encryption = 1;

  FormFactor = 2;

  HealthState = 5;

  InterconnectSpeed = 0;

  InterconnectType = 5;

  MaxMediaSize = 984816287;

  Name = "0_0_10";

  OperationalStatus = {2};

  RequestedState = 12;

  RPM = 4294967295;

  StatusDescriptions = {"OK"};

  SystemCreationClassName = "Clar_StorageSystem";

  SystemName = "CLARiiON+APM00134700308";

  TransitioningToState = 12;

};

I am IOPS can be calculated using RPM value. But the value looks relatively high(I am suspecting that it is # of sectors). Do you have any idea about this?

Thanks,

Aditya

[Paul von Behren]

4294967295 is the highest value you can store in a 32-bit field.  In other words, this is -1.  

RPM is not defined in SMI-S.  I assume it's something EMC added for internal use and my guess is that -1 means "unsupported.

As I mentioned before, I don't see how you can any useful measure ment of IOs-per-second from device instrumentation; IOPS are interesting when using actual measured values as seen from a server.  If the server has the SMI-S HBA profile, you can get data from from the HBA provider.  But you will get more useful results using the OS tools (such as Linux iostats).

--

[Aditya]

Thanks Paul. It solved the issue.

-

Aditya

To unsubscribe from this group and stop receiving emails from it, send an email to smi-s-developers-group+unsub...@googlegroups.com.