advice configure volume and pool on ME4

Hello geradmin,

You have 2 options to configure your ME4012 based on the information that you provided.

1. SSD RAID 10 = Pool A and Pool B has 7.2k in RAID x with 2x read cache drives shared.
2. SSD RAID10 and HDD RAID10 in tiering config, with data access shared by volumes in Pool A and Spinning disks with caching in Pool B.

You can’t have SSD tier and SSD Caching in same virtual pool as that is not allowed.  In addition to that having disks of same tier, in different raid group types, & in the same pool will make all drives operate at the slower RAID type, when accessing volumes that are spread amongst all disk groups in a virtual pool.

Please let us know if you have any other questions.

hi, thanks for reply. The scenario A was my fisrt choice, but what do you think about this configuration?

pool A -> SSD raid 1 + HDD raid 6 with one volume performance only and one volume shared

pool B -> SSD raid 1 +  SSD raid 1 + HDD raid 6 with one volume performance only and one volume shared and one volume archive only.

with modern controller like ME4 the raid 1 is much slower on writing that raid 10? and with 5 volume and tiering the controller can have problem with high workload?

thanks

Hi, i have another question. If i create a virtual volume with ssd and spinnig disk group, can i create a volume that only access to ssd group? on manual i find this

The three volume tier affinity settings are:

• No Affinity – This setting uses the highest available performing tiers first and only uses the Archive tier when space is exhausted in the other tiers. Volume data moves into higher performing tiers based on the frequency of access and available space in the tiers.
• Performance – This setting prioritizes volume data to the higher tiers of service. If no space is available, lower performing tier space is used. Volume data moves into higher performing tiers based on the frequency of access and available space in the tiers.

   

• Archive – This setting prioritizes the volume data to the lowest tier of service. Volume data can move to higher performing tiers based on the frequency of access and available space in the tiers.

So in my case i have a virtual pool with 4TB ssd abd 8TB hdd.

if i create a 3TB volume with performance tier and a second volume with the remain space (so 9TB) with no affinity i'm sure that that the 3TB are blocked exclusively by the first volume? and the second volume use 1 TB ssd and than 8TB hdd disks?

Hello geradmin,

If you have SSD drives and spinning drives in the same pool, your SSD will operate at your spinning drives speed.

Please let us know if you have any other questions.

hi, sure? i have configured a raid 10 ssd and raid 10 hdd in poolA, i have created a volume with no affinity. Today i ran the iometer and show on vmware metric 400K iop/s on sequential reading ( I hope that metric on vmware was right) . The virtual machine that i have tested was totaly on tier performace. i don't think that 7.2k rpm disks can have this performance.. can i check on san web gui? 

Hello geradmin,

Here is the link to the administrator guide & if you look on page 127 it shows you how to view the performance on your ME4012.  https://dell.to/38DyKz3

Please let us know if you have any other questions.

hi, thank you for reply. So can you definitively clarify how the san works with mixed disks?

so my test configuration is a 4 ssd raid10 and 4 hdd raid 10 disk group on same pool. one volume with affinity performance and one volume with no affinity.

in this configuration when i use the performance volume the controller use ssd speed? and when i use the other volume it use ssd or hdd if the data is cold or hot. right?

this is a iometer report from vmware machine connected to powervault by 10Gb rj45 cables

test disk on volume with no affinity

Test name Latency Avg iops Avg MBps cpu load

test disk on volume with tier performance

the specifications of test

'ACCESS SPECIFICATIONS =========================================================
'Access specification name,default assignment
Max Throughput-100%Read,ALL
'size,% of size,% reads,% random,delay,burst,align,reply
32768,100,100,0,0,1,0,0
'Access specification name,default assignment
RealLife-60%Rand-65%Read,ALL
'size,% of size,% reads,% random,delay,burst,align,reply
8192,100,65,60,0,1,0,0
'Access specification name,default assignment
Max Throughput-50%Read,ALL
'size,% of size,% reads,% random,delay,burst,align,reply
32768,100,50,0,0,1,0,0
'Access specification name,default assignment
Random-8k-70%Read,ALL
'size,% of size,% reads,% random,delay,burst,align,reply
8192,100,70,100,0,1,0,0

the strange is that the RealLife-60%Rand-65%Read in lower on volume performance. So there is something wrong on setup?

thank you

Hello geradmin,

When using performance tier affinity with mixed disk when data comes in, if it is random data will go to SSD first.  Any sequential data when it comes in will be written to highest tier of spinning drives first. Volume tier affinity is not the same thing as pinning, and it does not restrict data to a given tier and capacity.

Please let us know if you have any other questions.

ok, but in previous post you said that if i have two disk groups ssd and hdd on same virtual pool the ssd works at hdd's speed, right?

Hello geradmin,

That is correct.  When data is first written to a ME4 system with no data on it, data is first written to SSD.  Once the data is written to the SSD, then your ME4 will tier the data based on the profile that is used.  When you have data that is on both SSD and HDD, the data will be using HDD speed as it is the common speed for all drives in your configuration.

Please let us know if you have any other questions.

ok, " When you have data that is on both SSD and HDD" you mean that some chunk are on performance tier and other are on archive tier, right?

Hello geradmin,

That is correct. There can be times that data is on both tiers.  

Please let us know if you have any other questions.

Hi, the last week i have tried several raid combination. I found a strange thing about how raid controller works. So if in a virtual pool a create a raid 5 ssd and raid 10 HDD group and i create 2 volumes (one perfomance and one no affinity) the space allocation seems works right. But i create a raid 10 ssd and raid 10 hdd and keep the same volumes layout on vmware when a create a virtual disks (specialy zeroed eager) and i put data on volume performance the controlles allocate space on hdd raid and not on ssd. it's right? because I have several doubts about how it really works and manuals dosen't help at all.

thanks

Andrea

Hello Andrea,

As I explained earlier all random writes will go to your SSD drives.  All sequential writes will go to the next highest class of spinning disks.  There is not a way to make the data go to a single class of drives as you are trying to do.

Please let us know if you have any other questions.

ops.. i have missed this detail. So then in this real scenario how the controller works? i have a virtual linear pool with 2 volume with 2 affinity (performance and no affinity) like my case. I need to convert 2 hyper-v VM to vmware VM and put data in performance volume. The conversion will write sequential data and these data will be written on slower disk  if i understand correctly, right? . But what happen after? all data will be moved on performance tier after i will poweron the VM migrated? if not what should I do to migrate the VMs to the performance volume?

sorry, but I have to understand well before putting the infrastructure into production

thanks

Andrea