I was hoping to get a little more information about the Logical Capacity limits with XtremIO configs given what I have read in the education materials and deployment guides. My understanding is that the logical capacity/volume capacity shown in the dashboard is the amount of logical storage that has been configured as LUNs. Within this value's bar graph there is what is mapped out to the servers as LUNs and then what is actually consumed physically with unique data (feel free to drop some 411 on me if this is incorrect as it may impact the answers below ).
I have seen different values for the maximum logical capacity per cluster and trying to understand if this limit is a hard limit based on the architecture of each type of x-brick or really dependent on the type of data being stored given the deduplication/compression rates as noted in the release notes. For example, in one document I see 48TB as the max for a 20TB x-brick cluster and in the release notes this max value range from 73-130TiB. I would tend to see it as more of a range based upon the type of data being stored and the data reduction ratios achieved, but my questions are as follows:
1. Which is the real limit? Is it a hard limit or range as noted in the release notes?
2. Given the answer to #1 above, regardless of it being a hard or soft limit - what happens if the configured LUN space exceeds the limit (going by the definition that logical capacity is the amount of logical storage that has been carved up as LUNs)? Is there an alert or some sort or it is inferred that the system's physical capacity has met its finite limit?
Given there is an RPQ process for more than 130TiB, I'm just trying to understand what numbers are correct and what happens if they are exceeded?
Thanks for any insight!
The logical capacity limit that you mention above (48TB for a 20TB X-Brick) is based on an older code base (2.4.x and earlier).
With the 3.0.x code where we introduced inline Compression, the total maximum logical capacity limit is variable and depends on the data reduction (dedupe and compression) that you can get for the data that you store on XtremIO. It is not a hard limit as you can see based on the range that we specify in the release notes.
In order to help you size the array - you can assume that we will always be able to get you to at least 6:1 data reduction irrespective of the X-Brick capacity (assuming that the data can be deduplicated and/or compressed). I would recommend working with your EMC account team to get a better understanding of the space efficiency that XtremIO can deliver for your data. There are tools available that can scan your data and provide that level of insight.
Thanks Avi. I have seen that expected data reduction ratio and think I am OK on the concepts and not actually sizing an array to get the expected data reduction ratio. I just wanted to make sure from the previous codes with the hard limit that a customer wouldn't see any issues creating LUNs over 130TiB of total space given the range. I tested going over that and didn't see an issue so it sounds as expected....a variable recommended range to stay within versus any hard support limit. Doesn't seem like any alerts are triggered from the test I ran.
Thanks for the information. Feel free to add anything else if I have missed anything.
The logical capacity discussion above is related to the maximum amount of logical capacity that you can get out of an XtremIO array and not to the maximum volume size that you can create. Note that all volumes are thin provisioned in XtremIO. This means that you can create much larger volumes than the physical or logical capacity limit of the array. What you observed is the expected behavior of the array.
Thanks for the great info Avi. I was able to create larger than that so I saw the expected behavior. I just couldn't load that much data to get to the logical capacity range and that was what prompted me to see if there were any indicators built into the management around approaching or exceeding the range in the release notes since it is variable around the type of data that can be deduped/compressed. Let's say a customer by chance can exceed 130TiB logical capacity, given the note about the RPQ I just wanted to see if they would have trouble with support or get indicators warning them of that upper bound.
Yes, there are memory/metadata related alerts that would be sent out if you start reaching that limit. You can see a list of all the possible alerts and events in the XtremIO User Guide at XtremIO 3.0.1 and 3.0.2 Storage Array User Guide.
Hope this helps.