I have some doubts about setting Tiers in OCI. I have set some rules in volume tier quality like HDS AMS boxes should be Tier 2 and HDS R600/R700/USP/VSP are Tier 1 but when I generate reports, it shows that AMS boxes have both Tier1 and Tier2 disks.
Do I have to set different rule name for each vendor Or Different vendor setting will work under same rule?
After selecting the storage models in Tier edit section, I selected multiple options like disk type and speed. I wanted to know are that these multiple parameters used as AND or OR logic?
I also see some predefined entries in "Storage Tier Quality". Is it creating any conflict with my above settings?
It would be generous to characterize my feelings about the automatic tier annotation defaults in OCI as "doubtful". I suggest you carefully purge them.
First, go through the Annotation Rules and remove all of the rules that you or your customer did not create. Take note of what rules you've removed, and what objects they apply the tier annotation to.
Next, look at all of the objects to which unnecessary tier annotations may have been applied, find all of the unwanted tier annotations, and remove them. This will include, but may not be limited to:
Now your OCI database should be free of unwanted tier annotations.
Once you do an ETL, the Inventory Mart of your DWH will be free of them as well.
Cleaning up the other data marts is a tougher question. For all time-series entries prior to this point, the old, unwanted tier annotations will be present. You have two options:
Remember not to do reports using historical data that begins prior to this date
Reset your Data Warehouse database and start fresh from this date
Reset your Data Warehouse database and do a "build from history" and "init with current" to get your current annotations applied to some amount of history
When naming tiers, it may be a good idea to provide site-specific tier names. For instance, "Acme Tier 1" for customer Acme. This will make it easier to identify and destroy the default tier annotations.
I am testing this tier setup in my test environment. This is what I have done.
1. I cleared all defined tier annotations from all (internal volumes, pools, arrays, volumes)
2. Defined my own tier rules in storage pool
I can see the instant and right result in OCI Server UI, under volume details grouped by tier.
But problems are:
I cannot see the exact info in DWH reports which I see client UI.
I cannot find Tier asset/query item in inventory datamart (specially FC path, as my some reports are based on that). Although tier is available other datamarts like storage capacity. But tier item is one of the must have column which I need in my reports.
To see your new annotations in the DWH, first make sure you have completed an ETL since making the changes in the GUI. Second, for time-series data marts (ie everything but Inventory), make sure you are looking at the most recent time-series data.
To associate tiers with FC Paths in the Inventory mart, I believe you will need to join path (dwh_inventory.logical) to volume to object_to_annotation to annotation_value. A simplistic shot at this might look something like:
from volume, object_to_annotation, annotation_value
and volume.id = object_to_annotation.objectId
and object_to_annotation.annotationValueId = annotation_value.id
You'll want more than the volume name to properly identify volumes. This won't report any volume without an annotation, which may not be what you want. I think this will report only volumes that are directly annotated, not those that inherit an annotation from their storage pool or array, which may not be what you want. But hopefully you get the idea.
1. But I want to ensure that, Does "Simple Datamart" from all but inventory contains latest data and the advanced section of every datamart hold historical information.
If yes, then how come I see some arrays which were deleted from OCI server and ETL build many times?
2. I am trying to configure the tier rules for Netapp systems like I did for HDS and IBM. But due to some issue all available options (model, family, disk type, speed, size etc) ruled out. Is there any way to configure tier based on system name?
The arrays that have been inactivated and deleted from OCI prior to the last ETL should only appear in historical information in the DWH. Is there a document that leads you to believe that the simple marts do not contain any historical data? If so, can you point me to it?
I’m not aware of any way to parse storage array names to determine tier. This would potentially be a useful feature. You may want to talk to your SE about this, and ask him or her to open an RFE.
I haven't saw such document but based on the some report results and during my meetings with SE when asked they agreed that simple datamart are showing latest info. Infact I have shown some example reports to SE. Let me show here a simple example.
I am using "Storage and Storage Pool Capacity" datamart and I am just using storage name from "Physical capacity" in query studio. In that report output, I see names of storage systems which are inactivated and deleted from OCI server 2-3 weeks ago. ETL process is scheduled every midnight.
I was thinking to configure tier based on tier names because we have same models which are used for active-active clusters and Metro cluster, only thing which differentiate is their hostnames. I tried another thing. In OCI server, I manually selected some nodes -> right click -> set annotation -> Tier, and specified the teir value. But I do not see the same info in internal volume details when grouped by tier, it shows none.
I agree that the historical data that is showing up in simple data marts needs further looking into. Thank you for showing it to your SE.
For the tier annotations, confirm that all of the annotations you are using are being published to the DWH. From http://localhost/dwh on the DWH server, go to Annotations. Make sure you see a Tier annotation with a Storage target object, and that the Published box is checked. There may be others to publish here as well.
It is possible that the data mart only includes Tier annotations that are applied directly to internal volumes with its Internal Volume data. You would need to bring in the storage array’s tier annotation in that case.