• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Looking to replace Drobo datastore with something better

AMD_Gamer

Fully [H]
Joined
Jan 20, 2002
Messages
18,287
I spoke to my boss today about the Drobo and we are now talking about getting something to replace it. What do you suggest? I showed him the Synology DS1511+ and he said it was not a bad price. I could also suggest a ZFS whitebox.

These are the main VM's we will be running:

Server 2008 R2 (Primary DC)
Server 2008 R2 (Backup DC)
Server 2008 R2 (Exchange 2010/File Server)
Server 2008 R2 (SQL Server 2008 with low usage)
2-3 various other VM's
3-4 test Windows clients for testing stuff out before being deployed.

What would you suggest?
 
I simply don't do whitebox in production environments. Others can disagree, but that's me. As I've said in about every thread I love my Synology boxes. Just remember, they aren't an EMC box. If it goes down it'll be days to repair...not hours. If the Synology is the target budget then your choices are limited and nothing will be "enterprise" gear. So Synology, QNAP, and others. Or build something and support yourself. ;)

It's hard to make a solid recommendation without load criteria. Required IOPS? Read/Write percentages? What level of RAID protection do you need? Things like that. Then you can size the spindle count based on that. Else it's an educated guess.
 
Absolutely do not use a whitebox. Get something from a vendor on the HCL, otherwise you'll run into trouble even getting support from VMware when you call. A whitebox is for test labs at your house... I'm not even sure I'd use something like that for a lab at work.
 
For under $1K it's Synology, QNAP, whitebox, or something similar. But really...how much did the servers cost? And they only want to drop $1K on the device that actually holds, manages, and protects their data?
 
Yeah, I understand about "no whitebox in production". But as you say, if you're using something like synology, you are no worse off than a whitebox SAN and there you at least have some control over what it is doing, and how it does it. Of course, you are the only one who can really speak to the budget constraints (well, LOL, your boss...)
 
Trying to keep it under $1000.

Ouch.

As Netjunkie said, your options are very limited with that amount. You're basically looking at a small SAN comprised of SATA drives.

If you're getting by with what you have now, perhaps you should see if your boss can come up with more money as time goes by so you can look at more higher end products. EMC's VNXe series is a nice entry level SAN and starts at $10k.
 
Depending on the product from Synology you use, they are part of the VMware Ready Program, a whitebox ZFS server isn't. Though I love my ZFS box at home, I would never put one in work prod or lab.
 
Ok i will forget about a whitebox ZFS build lol.

Probably going to go with the Synology or QNAP.

I would also like a nice backup plan with this if we are going to spend money on it instead of relying on the raid 5/backup exec restore and having to spend a lot of time to recover.

I was just thinking about something like this: http://www.qnap.com/USEng/pro_detail_feature.asp?p_id=162 Get two of those 4 bay versions and have them mirror each other in case something bad happens to one of them? has anyone ever done something like that?

Also what do people think about Raid 5 vs Raid 01 for the datastore?
 
Synology can also mirror directories...but I've never tried it with VMware datastores. May be a problem since those files are always open...maybe I'll play with that and see what happens.

RAID5 vs RAID1/0 is really up to your needs for speed, recovery, and storage space.
 
Have you looked at a virtual san, something like StorMagic's SvSAN?
 
Synology can also mirror directories...but I've never tried it with VMware datastores. May be a problem since those files are always open...maybe I'll play with that and see what happens.

RAID5 vs RAID1/0 is really up to your needs for speed, recovery, and storage space.

It does not look like I will be going past the 2TB limit anyways so I think i will be switching over to RAID 01 for better recovery time.

And i did not necessarily mean mirror the two datastores. More like another datastore ready to go in the event the main one goes down without having to copy everything from my backup location.
 
It does not look like I will be going past the 2TB limit anyways so I think i will be switching over to RAID 01 for better recovery time.

And i did not necessarily mean mirror the two datastores. More like another datastore ready to go in the event the main one goes down without having to copy everything from my backup location.

There is no 2TB limit with NFS datastores. That's block level only.

So....how are you going to get the VMs to the second box? Mirror them, right?
 
$1000 including drives? Ouch that's low.

IMHO you would be much better off going with internal storage in those R515s if you are looking for the best performance out of your VMs for your budget. The only advantage to a SAN is scalable storage and multiple host access. Since you don't have a multi-host environment at the moment I personally would just fill my server with drives, and be done with it. Then in the future if you go to a multi host environment then purchase a real SAN.

Also if I remember right this is for an NPO. You may want to talk to Dell about pricing for a MD3200i. One with 4 450GB SAS6 15k drives, single controller, and 3 years of mission critical support is only $10k list, and should be more like $7-8k for partners / NPO. If you purchase it, and sell the drobo, you should be able to make back some of what you spent.
 
You can pick up an older Dell/EMC AX150i for about $3500 loaded with 250GB drives, less if you go with fewer. I originally picked up mine with 4x250GB and dual controllers for $1900, however I added the rest of the drives and the 1U cache battery. Which is WAY better than grabbing a QNAP or Synology.

We picked up our Dell MD3000i for $8,000~ I actually just had Dell sell us the licenses for Snapshot/Copy features, which was another $2k though. :(

At any rate, if you can do that small budget bump, you can get into a proper starter SAN to get your foot in the door; grab an AX150i if you can. :)

I've played with iSCSI on a QNAP, I was able to cause it to crash; that's all I can say about that... So unless you get one of the more expensive QNAP devices, it's not a good idea IMO. However once you get up to that price you can just go with an proper array.
 
Also if I remember right this is for an NPO. You may want to talk to Dell about pricing for a MD3200i. One with 4 450GB SAS6 15k drives, single controller, and 3 years of mission critical support is only $10k list, and should be more like $7-8k for partners / NPO. If you purchase it, and sell the drobo, you should be able to make back some of what you spent.

Heck, for $10k, you could get an EMC VNXe, NFS and iSCSI. Awesome integration with VMware, and let's face it..it's EMC, does it get any better?...lol..and no..I don't work for EMC....:D
 
$1000 with or without drives?

I mean honestly a Brand New Dell NX200 w/ 2TB Raid5 & 3yr warranty is only $1800 retail, NPO pricing would be less.
If you only want 1 yr warrant its $200 less.
 
Last edited:
So i was just thinking about this.

Get the Synology for the new iSCSI datastore and 4 2TB drives and use RAID 01. use the Drobo to replace our slow as molasses Buffalo Terastation Pro that we are using as our backup device with Backup Exec.

Now split the Drobo into 2 partitions. one partition will be RAID 5 for our normal backups with one of the Ethernet connectors attached to the normal network/vlan. The second partition and Ethernet connector will connect to my iSCSI VLAN and be used as backup in case the Synology goes down for whatever reason. I would do backups of all the VM's to this datastore so it would be ready to go in the event the Synology goes down.

What do you guys think about that? between that and the normal VM backups with Backup Exec I have setup that seems pretty solid?
 
$1000 including drives? Ouch that's low.

IMHO you would be much better off going with internal storage in those R515s if you are looking for the best performance out of your VMs for your budget. The only advantage to a SAN is scalable storage and multiple host access. Since you don't have a multi-host environment at the moment I personally would just fill my server with drives, and be done with it. Then in the future if you go to a multi host environment then purchase a real SAN.

Also if I remember right this is for an NPO. You may want to talk to Dell about pricing for a MD3200i. One with 4 450GB SAS6 15k drives, single controller, and 3 years of mission critical support is only $10k list, and should be more like $7-8k for partners / NPO. If you purchase it, and sell the drobo, you should be able to make back some of what you spent.

What do you mean i don't have a multi-host environment? I have two hosts and plan to implement HA when we are done moving everything.
 
With a SAN you don't need really care about ESXi failures and HA.
If one of your ESXi fails, you can just restart your VM on the second ESXi server within 5 minutes.

You have to care about a SAN failure and recovery time. While SAN HA is very expensive and
complicated, (beside backups) you should consider a second small SAN machine either with current
copies of your VM's to regain service within say 30 min.

For my own, i do this with blocklevel based ZFS replication to have a backup SAN with nearly realtime data.
Additionally i first try to move the real datapool to the second machine, if the first fails to restart original VM's.

Think about your maximal allowed time to regain running VM's
 
Right, because if a VM fails at 3 AM in the morning you'll restart it in 5 minutes. We call that "dreaming on". Granted that might not matter at this scale, but at that point you become responsible for the recovery and have to be worried about it all the time. How are you going to know there is a problem? Users copmlaining? God, I cannot even wrap my head around this stuff.
 
Well..do what you like but the answer is not always the one with the best benchmark results, especially when they are so close between the two. But...a Synology (and most of these small NAS boxes) won't really replicate iSCSI since the iSCSI LUN is a big binary file. You need to do that with NFS as it's just files on the filesystem..not a big blob.
 
Well..do what you like but the answer is not always the one with the best benchmark results, especially when they are so close between the two. But...a Synology (and most of these small NAS boxes) won't really replicate iSCSI since the iSCSI LUN is a big binary file. You need to do that with NFS as it's just files on the filesystem..not a big blob.

Well can I just use a 2TB volume and switch between iscsi/nfs to see which one performs better for me?
 
Let's be honest. With any of these boxes the protocol isn't going to be a bottleneck. It's going to be 4 or 5 7200RPM SATA drives. Go look at the VMware storage I/O whitepaper that compared NFS/iSCSI/FC. They are all VERY close in latency and IOPS. Enough not to matter unless throughput is a concern. That's why I don't do much iSCSI anymore except in specific cases.

Replication of iSCSI is more of a hassle on almost all NAS devices due to the fact that the NAS boxes see the iSCSI LUN as a giant file. A 2TB LUN on a NAS is a 2TB binary file. It's just not an elegant solution....and it's not just on these small boxes but on EMC and NetApp too (unless you do iSCSI to the back-end array on the EMC).

Remember. A NAS box is about serving up files from a filesystem. They pretend to be block devices with iSCSI.
 
We just got the Synology DS1511+ and 4X Western Digital RE4 WD2003FYYS 2TB drives for Raid 10. The GUI alone makes it better than the Drobo and its pathetic Drobo Dashboard :rolleyes:

Drobo will be for our backups.

I think i might talk my boss into selling the Drobo and maybe building our own backup NAS.
 
Let's be honest. With any of these boxes the protocol isn't going to be a bottleneck. It's going to be 4 or 5 7200RPM SATA drives. Go look at the VMware storage I/O whitepaper that compared NFS/iSCSI/FC. They are all VERY close in latency and IOPS. Enough not to matter unless throughput is a concern. That's why I don't do much iSCSI anymore except in specific cases.

Replication of iSCSI is more of a hassle on almost all NAS devices due to the fact that the NAS boxes see the iSCSI LUN as a giant file. A 2TB LUN on a NAS is a 2TB binary file. It's just not an elegant solution....and it's not just on these small boxes but on EMC and NetApp too (unless you do iSCSI to the back-end array on the EMC).

Remember. A NAS box is about serving up files from a filesystem. They pretend to be block devices with iSCSI.

I forgot to mention that i use an iSCSI Optimized switch. The Dell PowerConnect 5448. http://www.dell.com/us/enterprise/p/powerconnect-5448/pd?refid=pwcnt_5448

What do you think about this iSCSI optimization stuff? It sounds to me like it is just automatic QoS for iSCSI traffic? if you put iSCSI on its own VLAN with nothing else does the switch even do anything?
 
Be nice to find something in non Marketing BS that explains what iSCSI optimized L2 traffic means...
 
Be nice to find something in non Marketing BS that explains what iSCSI optimized L2 traffic means...

It just puts iSCSI traffic at a higher class of service by default.

As for putting iSCSI in its own VLAN...that doesn't really change anything. If the switch is congested it will treat all VLANs as the same unless you put one at a higher class. Again..why do iSCSI? Especially on a Synology? You're doing block-mode storage on top of a filesystem. Not efficient.

Just use NFS.
 
It just puts iSCSI traffic at a higher class of service by default.

As for putting iSCSI in its own VLAN...that doesn't really change anything. If the switch is congested it will treat all VLANs as the same unless you put one at a higher class. Again..why do iSCSI? Especially on a Synology? You're doing block-mode storage on top of a filesystem. Not efficient.

Just use NFS.

Ok thanks, I will try out NFS. iSCSI setup is a pain in ESXi if you ask me anyways.
 
Back
Top