Ask the Expert: Top Tips & Tricks to Rock Your XtremIO World

Hi Expert Panel,

Do all of the "Host-Side Best Practices" (HBA queue depth, Round Robin, etc) apply when a host running vSphere (Cisco UCS let's say) is zoned/connected to an XtremIO volume through VPLEX, as opposed to directly connected to an XtremIO array?

Hello Roberto,

I have a some general enquiries from customer. Basically, we are setting up an environment with XtremIO and VMware cluster servers. The software running on the VMware is Compuware which requires a 2.5Gb/s or higher throughput. for the 4 XBlock, what would be the consideration or best practice to follow to achieve this.

I have been searching best practices for XtremIO, however, the information I have so far are the top hints, host configuration guide, there have been no in depth whitepaper in regards performance tuning. Do we have anything from support or engineering?

thanks very much for your help.

Yixing Qiu

yixing.qiu@emc.com

If the host is connected through a VPLEX, you should be following the best practices recommended for VPLEX.

There is no performance tuning as such that you need to do on the array itself. A lot of that is already built in. Some of the high level performance data is available in the XtremIO data sheets.

Work with your XtremIO account team to discuss the workload details (r/w ratio, IO sizes, parallelism, etc) for more specific performance sizing.

GUI has an option to see the performance data going back to 2 years.

thanks Avi ! Would this be only from 4.0 ? and can you tell me from where ? I could not find where I can set ...

thanks !

Aya

Yes, this is applicable only from 4.0 release (and onwards). You should be able to see this option in the main dashboard (as a drop down list). I will try to send you a screenshot if you are not able to find it.

HI

About performance monitor , can we check last 2years record in version4 ? as release notes says below .  

Improved Data Collection and Reporting

Data is retained for up to two years and has variable granularity, according to the

  data age. The lowest sample granularity is 5 seconds.

If so , how can we set to check last 2years performance data ?

thanks !

Aya

hi ...

thanks for the reply !

like this ?

I found in event setting for 2years but it does not affect above performance date ....

So I am begging to think the records can be kept for 2years ,this is only for event /alerts ... Not for performance report ... ???

hi experts !

Is there any customer viewrable documents how to set PMI tunneling setting ? From4.0 version , there is change that management IPMI mode are no longer supported .. there is only IPMI connection via SP  .( Only  Support for IPMI connection via SP )

thanks !

Aya

We do maintain performance history also for up to two years.

The main dashboard will not show the two years of performance history. However, you can see that by exporting the performance stats using the export-performance-history command.

1. 1. You cannot install new components inside the vXMS. Note that the array would not stop data processing even though you lose the XMS for any reason. Typically, customers have used VMware DRS to protect the vXMS – just like you would protect any other VM.

2. 2. Typically, we recommend using RestAPI for automation scripts. It has much more flexibility and a commitment from our side to maintain backward compatibility. Your scripts based on CLI might break if there is a change in the output structure in one of the future releases.

3. 3. The cluster would internally generate a PIN and this is never exposed through any user interface. As you noted, if there is a concern with changing it – we can always re-encrypt it and the cluster will generate a new key.

4. 4. There is a VSPEX reference architecture based on XenServer (and many other collaterals). Check XtremIO.com for the most up to date collaterals. Or your EMC account team should have these links.

As a proud owner of a 1 day old XtremIO installation i would like to ask some "operational" questions

1) How to protect vXMS, can i back it up with Avamar ?  ( I do see that VMware tools are not installed)

2) Output format from xmcli commands - let's just say that there has to be a way to specify which columns you want to display.  Unless you have a monitor that supports 3840x2160 resolution, columns wrap around and it's impossible to read anything.  Am i the first the complain ?

3) Encryption PIN - I was sitting next to my ASE during install and i don't remember being prompted to enter a phrase or anything like that. So what is the PIN and how is it derived (i understand it can be changed even with customer data on the cluster)

4) I don't see any best practices/recommendations for Xen Server environment, anything you have to share ?

Thank you

Hi Avi,

1) VMware HA is not backups, you and I know that. My vXMS resides on a VNX LUN, in an unlikely VNX pool corruption i would like to recover my VM and and not start from scratch, loosing my historical performance data.  Really, no one has asked to protect vXMS ?

2) So the answer is no, we have to deal with this clunky, unreadable output. Why even offer it ?

3) I wanted to make sure this was not based on a pass-phrase (like in Data Domain encryption for example), where i provide the string that can be used to lock/unlock encryption.

4) Couldn't find much online but some old references on Itzikr's blog. I am looking for something official.

Thanks