HDS HUS-VM vs. NetApp 6220

[BigBjorn]

Hi all,

We are in the process of upgrading our current NetApp 3140 MetroCluster solution. We are out of I/O and diskspace also controllers are running out of both CPU and RAM.

Currently we are looking on a few players out there and try to have an open mind. HDS is pretty unknown to us.

Our NetApp in MetroCluster and NFS in an VMware environment has been working very well management and DR-wise. HDS HUS-VM will offer same functionality according to HDS, and the design looks pretty good.

For you who have experience from both from a VMware perspective: What is the good, the bad and the ugly?

Specs:

FAS 6220 in MetroCluster

SAS drives with SSD cache

FlashCache (accelerating our older FC disks that we are keeping)

OnCommand Balance

vs.

HDS HNAS3080 with HUS-VM

SAS 10k drives (No acceleration, but higher number of drives compaired)

Hitachi CommandSuite + Tuning Manager

Thin info about specs, but the topic is more about how my life as an VMware admin will be with HDS or NetApp.

THanks

/B

[vangoose]

I've used both platforms for VMware infrastructure, if you ask me, I wouldn't touch HNAS in a near future for VMware using NFS.

HNAS is in fact a good product, feature and spec look good, but it's optimized for different king of workload, more on throughput rather than latency.

The company I worked for migrated from NetApp to HNAS and it turned out to be a disaster. We had TEST/DEV environment on NetApp using single SATA shelf. The same workload after moving to HNAS requires 90 SAS disks to handle and still has twice the latency, and still close to useless. 2 years later they decided to move all the workload to backend HDS FC SAN from HNAS because even with HDS's help, the problem could not be resolved. If my memory still works, the HNAS model I had was 3090 and with hundreds 15K SAS disks in the back, the latency is at 100ms over 10Gb.

I manage a lot of NetApp at current job, 6280 is used in our datacenter and I can't be happier. Also don't forget you will loose things like SMVI if you move to HNAS. HDS did provide us with scripts to do things very similar to SMVI that we integrated with Netbackup but you don't get nice GUI from VSC.

[BigBjorn]

Thank you for your input!

100 ms sounds like a error of some kind. What did HDS say about that?

How was management compaired?

Thanks

[vangoose]

HDS initially said it was a bug but version after version, they couldn't resolve the issue. For me, it really means HNAS is not designed for VMware workload which is highly random IO with low latency. Don't believe their PR, 2M IOPS means nothing if latency is 100ms.

Management is another thing, with SAN + NAS gateway model, you will have to manage 2 different components. NetApp is truly unified, and you can use VSC to provision or OnCommand/command/powershell to create/export volumes and mount on esx.

BTW. back in 2 years when we deployed HNAS, there was 128 file system limit. Not sure where they stand today.

[xattila]

Vangoose

Little surprised at other response but here is current information. This is a reference architecture on using current generation HDS Enterprise NFS platform with VMware http://www.hds.com/assets/pdf/deploy-hitachi-ucp-select-for-vmware-vsphere-using-hnas.pdf

HNAS is a rockstar for VMware NFS environments and even delivers lower latency than Brand X FC SAN based environment based on VMmark testing as it uses hardware accelerated filesystem. It thrives on random I/O as effectively you have four controllers in a cluster with some secret sauce for sequentializing random I/O.... Plus, as a VMware admin, it has all the native vCenter adapters plus a VSC like vCenter plugin called Virtual Infrastructure Integrator (V2I).Hitachi NAS V2I_Video - YouTube

Brand N is decent classic filer but there are formidable reasons why we are seen upgrades to an Enterprise NFS platform. I'll let you experience the experience and get off the soap box !

disclaimer: I work for HDS

[Josh26]

xattila wrote:

http://www.hds.com/assets/pdf/deploy-hitachi-ucp-select-for-vmware-vsphere-using-hnas.pdf

HNAS is a rockstar for VMware NFS environments

So says the person that bumped a two month old thread to promote a product.

[doctormiru]

Hi BigBjorn

Your post is almost one year old meanwhile. But I'm interestedwhat you've choosen for you infrastructure, then I'm in the same situation as you was one year before. We've a NetApp FAS3240 Metrocluster and we'll be out of resources in a short time. So, I know the advantages of a NetApp storagesystem together with vmwware. We've configured all VMware datastores as NFS datastores and for some VM's direct attached iSCSI LUN's into the vm (for exchange or mssql for example). We have to plan a replacement/upgrade of this infrastructure now. One way is to continue with NetApp, but I'm also interested in other storage solutions, as they give me an adequate solution.

I think, with the MetroCluster thing, NetApp is the only vendor on market. Or, are there other vendors with the same highavailability feature? Is the HDS HNAS comparable with a NetApp system, if it's used as VMware storage (filebased, not blockbased storage). The backup topic we've addressed at the moment with VSC for a snapshot backup onto the primary system and SnapVault technology to backup the VMware volumes onto a secondary storage system (not VSC backup, crash consistent backup only).

Can you tell me, if you've found "the best" solution for you?

Thanks