VMware vSphere

Hi All,

I've been looking at the DroboPro to get us started with our new app (with plans to expand to something else a bit more scalable). Has anyone tried using ESXi 4 Update 1 with the DroboPro to see if it has improved the writing throughput issue people were reporting?

Thank you!

-wavesound

Bump.

Sorry to see you've not gotten a response. We're considering the new Drobo Elite for it's great price/storage value. Apparently it's VMWare approved. Heard anything about this new model? Compared to other SAN's the price would allow using a couple of them for redundancy.

I'll be installing a Drobo Elite tomorrow. I'll be running several file server VM's against it, and using it for archive space. I have Tier 1 storage, but at $30K for a 5TB node that gives you 2.5 TB I can't justify getting another one at the moment. 16TB for $6K is a fantastic price. I'll post again once we start using it.

I just installed Drobo Elite with 2 ESX 4.0 Update 1. At the first few days, it was working perfectly. I was able to perform Storage VMotion on my vms. However, lately, it started to act up. Especially, when I was performing Storage VMotion. The Storage VMotion would stop and cause all my vms to stop responding. I checked the vmkernel logs and found all my datastores are timed out. I had to reboot Drobo Elite in order to bring them back. It is very frustrating. I've been communicating with both VMware support and Drobo support, but so far, there is no solution.

I also am having the same issues with DroboElite. I really wish someone could give a little insight into why it isn't working at all. I have 3 powered on virtual machines. 1 DB, 1 Web and 1 File, we want to have 6 at minimum powered on at a time. Is this too much to ask from droboelite or iscsi?

The problem is not with iSCSI, but with the Elite. At this point, I wish I never bought it. If you use it to create RDM's with NTFS partitions, it works fine. Once you put a VMFS datastore on it, expect to see 10 to 20MB/sec throughput. Do not put more than one VM per datastore. I would guess no more than three VM's on the Elite overall. DataRobotics kept blaming performance on my setup and VMware, then they put out a patch that helped a little with write performance. For every problem I've had with the Elite, DataRobotics has tried to blame someone else, and in the end the problems keep coming back to them. Stay away from this product if your wanting to store VM's on it. If you need it for RDM storage, then it may work for you.

I've experienced almost every issue brought up in this forum and Data Robotics is giving me the same runaround. They refuse to admit fault and continue to direct me back to VMWare. I asked them if they can help facilitate a return with the retailer where it was purchased but they would not. Now we're stuck with a box that barely good at much of anything. There are NAS's that I can buy for a few hundred bucks that have much better throughput than this thing.

Just a note for those with a DroboPro (or looking to get one), and vSphere 4.

We have been having trouble with our DroboPro losing connection & hanging when being accessed through a vm guest. Everything appeared to be setup and configured fine, the Drobo shows up correctly as a iSCSI target and we are able to map vm disks to it. Initially, this seemed to work fine, but over the past week or so we have found the ESX4 host is consistently dropping connection when the guest is accessing the Drobo.

Following is the message I sent to them...

We have connected our DroboPro as iscsi to our esx4 host and have been experiencing connection issues (frequent disconnects) when accessing data from the vm disk on the drobo. I noticed in the documentation that ESX3.5 is 'officially' supported for the droboPro, but saw no mention of esx4. Is this a known problem with the droboPro on esx4 hosts? Do you have any documentation or information that might help us resolve this issue?

...and their response...

Hello Josh,

Thank you for contacting Data Robotics Technical Support.

At this time DroboPro does not work with vSphere, only the DroboElite.

If anyone has any ideas or input I'd love to hear it... otherwise looks like I'm on my own. Good luck everyone.

I used to have a Drobo Pro and had the same problem as you. Luckily, I was able to have it replaced with Drobo Elite from the reseller. However, Drobo Elite have the same problem whenever I perform "hot" Storage VMotion on my vms. It will drop connection and cause all my vms to freeze. I was able to pull out diagnostic data and contacted their level 3 tech support, but they said that everything looks normal. At this point, I am able to use it as long as I don't perform "hot" Storage VMotion which is a bummer. Drobo has caused me so much headache and I wouldn't recommend it to anyone for their VMware infrastructure.

Hi All,

The DroboPro has been certified with VMware 3.5 only and not with vSphere 4 and the DroboPro is really focused as a single server, direct attached storage system so if you have more than one ESX server you want to use the DroboElite.

The disconnecting or slow performance of a DroboElite can be symptomatic of incorrect OS alignment within the VMFS datastore. This is particularly the case with Windows Server 2K3, WinXP, and Linux VMs. This is a common problem amongst many storage systems and is a characteristic of VMware and not the DroboElite in particular. VMware has documented some knowledge base articles and best practices around the OS alignment issue. Check out these links from VMware for more info...

http://www.vmware.com/pdf/esx3_partition_align.pdf

http://www.vmware.com/pdf/Perf_Best_Practices_vSphere4.0.pdf

http://www.vmware.com/files/pdf/vmfs_rdm_perf.pdf

Even though the first document is on ESX 3 there is still very relevant information for vSphere.

For best performance on the DroboElite with VMware it is critical that OS alignment is optimized.

As always, support for Drobo products can always be provided by the Data Robotics Support Team at 1.866.426.4280.

The most amazing thing to me is how the DroboPro ever got certified. Did you know that while it is running storage for ESX, you can't tell what's going on in the DroboPro? It doesn't support hooking up an admin machine while it is connected to ESX, and all the management and monitoring is done from the dashboard on a client PC - not from the DroboPro itself.

Further, if you want to add, change, or delete a volume, you need to shut down all VMs, shut down all ESX Servers connected to the DroboPro, then hook up a USB cable to an admin laptop, make the changes you want, then disconnec the USB and start up the ESX Servers and VMs again. Once it is online, the only way to get very limited status is to hang a webcam in front of the DroboPro and look at the pretty lights.

How could something like this ever get certified? Makes me wonder what kind of criteria is used.