i would like to know different opinions when we allocate storage to ESX cluster.
allocating from a LUN from a storage pool over a raid group ( 6+2 ), which one will give more performance for my ESX considering a mix load of R/W.
As noted above, an answer to your question would require more information.
Many customers use HDP or HDT pools, and leverage the in-array tools to place and migrate data according to SLA, Value, and $$. What is the Goal? Best Performance? Lowest Cost? Maximum Scalability of the Array? Least amount of monitoring and changes? Depending on your Goal, there would be different recommendations.
There are a few discussions already in the performance and sizing forum, best practice and deployment guides are also available on Customer Portal. In general I see a lot of customers choose smaller LUNs with fewer VM's per LUN. This allows them the highest degree of flexibility to move, migrate, replicate, identify bad actors, etc... They initially allocate the LUN on the tier that "best fits" the expected workload profile and demand of the customer. As we all know reality can be wildly different. Other customers leverage large HDT pools to allow the workloads to seek their best automatic demand fit.
the storage pool will provide better performance if configured as per best practices. Some customer testimonials can be viewed here http://www.techvalidate.com/product-research/hds-virtual-storage-platform/case-studies
To be honest, there is no straight forward answer to this but if you carefully look at some of the important points and take not only the pool or RG in picture but things like qdepth, no of VM per Lun/Lun size/Work load /RG type and the list will go on...all are inter related.
In pools, the striping is very beneficial; however, one has to make sure that you don't assign too large luns in few numbers, which can cause perf issues due to high qdepth. I think with pools the IO workload can be distributed much efficiently but as I said you need to take care about other things which are interdependent.
Hope this helps ..
One of our customer uses LDEV's from DP Pool with a standard size of 1024 GB per LDEV.
And like others advise, there is no such global recommendations. that is purely based on your requirement and in which environment you are going to deploy.
Retrieving data ...